diff --git a/docs/ebsd-parameter-selection.md b/docs/ebsd-parameter-selection.md new file mode 100644 index 0000000..6c96174 --- /dev/null +++ b/docs/ebsd-parameter-selection.md @@ -0,0 +1,152 @@ +# How EBSD Parameters Are Decided + +This note explains *why* the acquisition parameters recommended for Electron +Backscatter Diffraction (EBSD) take the values they do — the physics and +statistics that fixed the settled optima recorded in standards such as +ASTM E2627 and ISO 13067, rather than simply pointing at those standards. + +Every EBSD parameter is a point on an explicit trade-off curve balancing +**pattern signal/contrast** vs. **spatial resolution** vs. **angular precision** +vs. **acquisition speed** vs. **statistical representativeness**. The underlying +physics (backscatter yield, interaction volume, `λ ∝ 1/√V`, `SNR ∝ √N`, gnomonic +projection geometry) fixes the *shape* of each curve; sampling and confidence +statistics fix *how far along* it you need to go. + +> Note on provenance: the reasoning below is derived from established EBSD +> physics and metrology. It was **not** produced via Edison Scientific — that +> path is currently blocked by an empty `EDISON_PLATFORM_API_KEY` in CI. Once a +> non-empty key is provided, the same query can be re-run through the +> `LITERATURE_HIGH` (paperqa3-high) job to produce a fully citation-grounded +> version, and the trajectory artifacts committed alongside this document. +> Verify the exact numeric clauses of ASTM E2627 / ISO 13067 against current +> published editions before using them for compliance. + +## 1. Specimen tilt ≈ 70° + +The backscatter coefficient η rises steeply with tilt, and — more importantly — +the *fraction* of backscattered electrons that retain coherent diffraction +information (the Kikuchi signal, versus diffuse background) peaks in roughly the +60–75° range. Tilting shortens the escape depth normal to the surface, so the +emerging electrons sample a thin near-surface layer where their exit +trajectories still carry channeling/diffraction contrast. Above ~70° the pattern +contrast keeps improving marginally, but the projected beam footprint elongates +(spatial resolution becomes strongly anisotropic), geometric +foreshortening/distortion grows, and shadowing worsens. **70° is the empirical +optimum** balancing Kikuchi-band contrast against spatial resolution and +geometric distortion — that is why the standards fixed on it, not the reverse. + +## 2. Accelerating voltage (typ. 10–30 kV, 20 kV default) + +Two competing physics chains set this: + +- **Signal / pattern sharpness:** electron wavelength `λ ∝ 1/√V`, and Kikuchi + band width scales as `~2λ/d` (Bragg angle). Higher kV → shorter λ → thinner, + sharper bands and higher BSE yield → better statistics and, if the detector + can resolve the thinner bands, better angular precision. +- **Spatial resolution:** higher kV enlarges the interaction/excitation volume + (deeper, wider), degrading resolution and blurring measurements near + boundaries. + +So 20 kV is a general-purpose compromise; you drop to ~10 kV for nanoscale +features, thin films, or beam-sensitive/low-Z materials (smaller interaction +volume), and push to 30 kV for weakly scattering light-element phases that need +more signal. Low-Z elements have low BSE yield → need higher kV/current; high-Z +phases scatter strongly → tolerate lower kV. + +## 3. Beam / probe current + +Pattern SNR `∝ √N` (electrons per pattern), so more current gives cleaner +patterns and shorter exposure (faster maps). The cost is probe size: for a given +column, higher current means a larger probe → worse spatial resolution (less +severe on FEG than thermionic sources). You raise current for speed/SNR on +coarse microstructures and lower it (small probe) to resolve fine grains and +boundaries. + +## 4. Step size — the "≥ ~10 points per grain" rule + +This comes from sampling statistics, not convention. If a grain is covered by N +measurement points, the fractional error in its reconstructed area/diameter from +pixel discretization falls roughly as `~1/√N`, and boundary/grain-count +reconstruction becomes unreliable when the smallest grains contain only a +handful of points. Choosing step ≤ (smallest grain of interest)/10 puts ≥ ~10 +points across a grain, bringing pixelation error down to a few percent and +letting a grain-reconstruction algorithm (misorientation-threshold flood fill, +typ. 5°) distinguish real grains from noise. **ASTM E2627** codifies exactly this +logic (minimum points-per-grain and a misorientation threshold for defining a +grain) so that grain-size measurements are reproducible. The lower bound on step +size is physical: nothing is gained by stepping below the effective spatial +resolution (~20–50 nm for FEG at 20 kV) or below what intragranular-gradient +(GND) sensitivity requires. + +## 5. Working distance, detector distance, and pattern centre + +WD (~15–20 mm) is set so the tilted region of interest projects onto the phosphor +with correct geometry. Detector insertion distance trades **angular capture** +against **gnomonic magnification**: a closer detector captures a wider slice of +the diffraction sphere (more bands → more over-determined, more robust indexing) +but lower magnification per band. The **pattern centre** must be calibrated +because indexing solves orientation from band positions *relative to the PC* — a +PC error maps almost linearly into an orientation error, which is why +calibration (known-orientation standard, moving-screen, or iterative refinement) +is mandatory rather than optional. + +## 6. Camera binning, exposure, gain + +Detector pixel count sets the ceiling on angular resolution: band-position +precision improves with more pixels across the pattern, so orientation precision +does too — but at lower per-pixel SNR and slower readout. **Binning** (4×4, 8×8) +sums pixels for higher SNR and frame rate at the cost of angular resolution: +heavy binning for fast mapping, low/no binning for high-precision or HR-EBSD +(resolving <0.1° misorientations). Exposure/frame-averaging improves SNR as +`~√(frames)`, sharpening Hough peaks and indexing reliability; you set it to the +minimum that reliably meets your indexing-quality target (CI/MAD) at the desired +speed. Gain amplifies signal *and* noise, so it doesn't substitute for real +exposure. + +## 7. Hough transform parameters + +More detected bands over-determine the orientation fit → higher precision and +fewer misindexes, with diminishing returns and slower indexing (typ. 7–12 +bands). Finer Hough (θ, ρ) binning improves band-position precision (→ better +angular resolution) but is noisier and slower; the butterfly-mask size is tuned +to the expected band width, which itself depends on kV and detector distance. +These are chosen to sit at the knee of the speed / indexing-success / precision +trade-off for the material. + +## 8. Minimum grains / scan area for representative statistics + +The counts in standards are confidence-interval arithmetic. The standard error +of the mean grain size `∝ σ/√(n_grains)`, so to bound the mean to ±X% at 95% +confidence you need `n ≥ (1.96·σ / (X·mean))²` grains — a few hundred to a few +thousand depending on size spread. Texture (ODF) representativeness is stronger +still (typically thousands of grains for a stable ODF, more for weak/random +textures). The required scan area then follows from +(grains needed) × (mean grain area), subject to the step-size constraint — which +is what ultimately sets map size and acquisition time. + +## 9. CI / MAD / band-contrast thresholds for clean-up + +These cutoffs were set empirically by correlating each metric against +independently confirmed correct indexing: + +- **Confidence Index** (TSL) `= (V₁−V₂)/V_ideal` from the band-voting scheme; + **CI > 0.1** is the conventional "reliable" line because, across many + materials, >99% of points above it index correctly. +- **MAD** (mean angular deviation, Oxford) is the mean misfit between detected + and simulated bands; **MAD < ~1°** marks a good fit. +- **Band-contrast / image-quality** thresholds separate poorly diffracting + points (boundaries, deformed regions, unindexed phases). + +Clean-up routines (neighbour-orientation correlation, grain dilation) use these +thresholds to drop/replace unreliable points — and standards require reporting +the clean-up because over-aggressive filtering biases grain-size and texture +results. + +## Summary + +The through-line is that every EBSD parameter sits on a physics- and +statistics-defined trade-off curve. The physics fixes the shape of the curve; +sampling and confidence statistics fix how far along it you must go. The +standards simply record the settled optima of those curves for common materials +— which is exactly the reasoning CALIBER aims to make explicit and +sample-specific when recommending baseline acquisition parameters. diff --git a/docs/images/convention-validation.png b/docs/images/convention-validation.png new file mode 100644 index 0000000..e73d2ee Binary files /dev/null and b/docs/images/convention-validation.png differ diff --git a/docs/images/euler-angles.png b/docs/images/euler-angles.png new file mode 100644 index 0000000..129ae50 Binary files /dev/null and b/docs/images/euler-angles.png differ diff --git a/docs/images/eutectic-si-schematic.png b/docs/images/eutectic-si-schematic.png new file mode 100644 index 0000000..c26f94f Binary files /dev/null and b/docs/images/eutectic-si-schematic.png differ diff --git a/docs/images/gnomonic-projection.png b/docs/images/gnomonic-projection.png new file mode 100644 index 0000000..afe62f4 Binary files /dev/null and b/docs/images/gnomonic-projection.png differ diff --git a/example/area2-results/area2_full.ang.gz b/example/area2-results/area2_full.ang.gz new file mode 100644 index 0000000..44d9c96 Binary files /dev/null and b/example/area2-results/area2_full.ang.gz differ diff --git a/example/area2-results/ci_hist.png b/example/area2-results/ci_hist.png new file mode 100644 index 0000000..09397d8 Binary files /dev/null and b/example/area2-results/ci_hist.png differ diff --git a/example/area2-results/dictionary-indexing/area2_slice_di_refined.ang.gz b/example/area2-results/dictionary-indexing/area2_slice_di_refined.ang.gz new file mode 100644 index 0000000..6e23cef Binary files /dev/null and b/example/area2-results/dictionary-indexing/area2_slice_di_refined.ang.gz differ diff --git a/example/area2-results/dictionary-indexing/di_quality_maps.png b/example/area2-results/dictionary-indexing/di_quality_maps.png new file mode 100644 index 0000000..75f61e1 Binary files /dev/null and b/example/area2-results/dictionary-indexing/di_quality_maps.png differ diff --git a/example/area2-results/dictionary-indexing/ipf_hough_vs_di.png b/example/area2-results/dictionary-indexing/ipf_hough_vs_di.png new file mode 100644 index 0000000..ae8d57b Binary files /dev/null and b/example/area2-results/dictionary-indexing/ipf_hough_vs_di.png differ diff --git a/example/area2-results/dictionary-indexing/ncc_histograms.png b/example/area2-results/dictionary-indexing/ncc_histograms.png new file mode 100644 index 0000000..d5fa0a2 Binary files /dev/null and b/example/area2-results/dictionary-indexing/ncc_histograms.png differ diff --git a/example/area2-results/dictionary-indexing/pattern_vs_simulation.png b/example/area2-results/dictionary-indexing/pattern_vs_simulation.png new file mode 100644 index 0000000..8f47aa7 Binary files /dev/null and b/example/area2-results/dictionary-indexing/pattern_vs_simulation.png differ diff --git a/example/area2-results/dictionary-indexing/stats.json b/example/area2-results/dictionary-indexing/stats.json new file mode 100644 index 0000000..adcd9a0 --- /dev/null +++ b/example/area2-results/dictionary-indexing/stats.json @@ -0,0 +1,36 @@ +{ + "n_patterns": 4900, + "dictionary_size": 52607, + "dictionary_resolution_deg": 2.5, + "pc_bruker": [ + 0.531, + 0.3204, + 0.7338 + ], + "hough": { + "s_per_map": 10.6, + "ci_gt_0.1": 0.9927, + "median_ci": 0.672, + "median_fit_deg": 0.514 + }, + "di_al": { + "s_projection": 13.0, + "s_indexing": 48.4, + "s_refinement": 87.3, + "median_ncc_di": 0.221, + "median_ncc_refined": 0.285 + }, + "agreement_hough_vs_di": { + "median_misorientation_deg": 0.78, + "frac_within_2deg": 0.794, + "frac_within_5deg": 0.831, + "median_misori_where_ci_lt_0.1": 41.42, + "n_ci_lt_0.1": 36 + }, + "phase_test_al_vs_si": { + "median_ncc_al": 0.285, + "median_ncc_si": 0.209, + "frac_al_wins": 0.991, + "median_delta_ncc": 0.073 + } +} \ No newline at end of file diff --git a/example/area2-results/grain-reconstruction/stats.json b/example/area2-results/grain-reconstruction/stats.json new file mode 100644 index 0000000..16d9608 --- /dev/null +++ b/example/area2-results/grain-reconstruction/stats.json @@ -0,0 +1,216 @@ +{ + "baseline": { + "config": { + "mis_tol": 5.0, + "min_size": 10, + "fit_max": 1.4 + }, + "n_raw_grains": 10885, + "n_grains": 1905, + "retained_frac": 0.9870733046581861, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 14.177289635700411, + "area_weighted_mean_diam_um": 18.989895851086057, + "speckle_grain_frac": 0.0, + "largest_grain_px": 544 + }, + "tol_1deg": { + "config": { + "mis_tol": 1.0, + "min_size": 10, + "fit_max": 1.4 + }, + "n_raw_grains": 34274, + "n_grains": 1426, + "retained_frac": 0.9246476617552851, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 16.298211698524327, + "area_weighted_mean_diam_um": 19.244197904489038, + "speckle_grain_frac": 0.0, + "largest_grain_px": 306 + }, + "tol_2deg": { + "config": { + "mis_tol": 2.0, + "min_size": 10, + "fit_max": 1.4 + }, + "n_raw_grains": 14711, + "n_grains": 1930, + "retained_frac": 0.9774412006955249, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 14.164493528234063, + "area_weighted_mean_diam_um": 18.11048188186429, + "speckle_grain_frac": 0.0, + "largest_grain_px": 513 + }, + "tol_15deg": { + "config": { + "mis_tol": 15.0, + "min_size": 10, + "fit_max": 1.4 + }, + "n_raw_grains": 8761, + "n_grains": 1583, + "retained_frac": 0.9903564564839389, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 15.035941200927164, + "area_weighted_mean_diam_um": 23.863433998664373, + "speckle_grain_frac": 0.0, + "largest_grain_px": 1391 + }, + "min_1px": { + "config": { + "mis_tol": 5.0, + "min_size": 1, + "fit_max": 1.4 + }, + "n_raw_grains": 10885, + "n_grains": 10885, + "retained_frac": 0.9965223757664501, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 4.705369550960168, + "area_weighted_mean_diam_um": 14.680316447908583, + "speckle_grain_frac": 0.8249885163068443, + "largest_grain_px": 493 + }, + "min_100px": { + "config": { + "mis_tol": 5.0, + "min_size": 100, + "fit_max": 1.4 + }, + "n_raw_grains": 10885, + "n_grains": 109, + "retained_frac": 0.6348837741374577, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 49.6782286968591, + "area_weighted_mean_diam_um": 54.63153849288807, + "speckle_grain_frac": 0.0, + "largest_grain_px": 1248 + }, + "min_400px": { + "config": { + "mis_tol": 5.0, + "min_size": 400, + "fit_max": 1.4 + }, + "n_raw_grains": 10885, + "n_grains": 2, + "retained_frac": 0.0498192550562826, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 105.11354280066635, + "area_weighted_mean_diam_um": 105.89189226227053, + "speckle_grain_frac": 0.0, + "largest_grain_px": 2442 + }, + "filter_none": { + "config": { + "mis_tol": 5.0, + "min_size": 10, + "fit_max": "off" + }, + "n_raw_grains": 11127, + "n_grains": 1907, + "retained_frac": 1.0, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 14.265263807938679, + "area_weighted_mean_diam_um": 19.07423004832693, + "speckle_grain_frac": 0.0, + "largest_grain_px": 544 + }, + "filter_strict": { + "config": { + "mis_tol": 5.0, + "min_size": 10, + "fit_max": 0.5 + }, + "n_raw_grains": 12404, + "n_grains": 955, + "retained_frac": 0.2633728379244074, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 10.538771717393518, + "area_weighted_mean_diam_um": 13.124922839829972, + "speckle_grain_frac": 0.0, + "largest_grain_px": 149 + }, + "strict_erode2": { + "config": { + "mis_tol": 5.0, + "min_size": 10, + "fit_max": 0.5, + "erode_iters": 2 + }, + "n_raw_grains": 12404, + "n_grains": 955, + "retained_frac": 0.6201038711448705, + "cleanup_altered_frac": 0.356731033220463, + "mean_diam_um": 16.48344210121238, + "area_weighted_mean_diam_um": 18.83655208768471, + "speckle_grain_frac": 0.0, + "largest_grain_px": 278 + }, + "strict_erode10": { + "config": { + "mis_tol": 5.0, + "min_size": 10, + "fit_max": 0.5, + "erode_iters": 10 + }, + "n_raw_grains": 12404, + "n_grains": 955, + "retained_frac": 0.9787338702297063, + "cleanup_altered_frac": 0.7153610323052988, + "mean_diam_um": 20.51236453396346, + "area_weighted_mean_diam_um": 24.016520820198195, + "speckle_grain_frac": 0.0, + "largest_grain_px": 387 + }, + "strict_fill": { + "config": { + "mis_tol": 5.0, + "min_size": 10, + "fit_max": 0.5, + "fill_defect": 1000 + }, + "n_raw_grains": 12404, + "n_grains": 955, + "retained_frac": 0.2917429303559989, + "cleanup_altered_frac": 0.028370092431591387, + "mean_diam_um": 11.023068771429246, + "area_weighted_mean_diam_um": 14.062454513503473, + "speckle_grain_frac": 0.0, + "largest_grain_px": 173 + }, + "worst": { + "config": { + "mis_tol": 1.0, + "min_size": 1, + "fit_max": "off" + }, + "n_raw_grains": 34548, + "n_grains": 34548, + "retained_frac": 1.0, + "cleanup_altered_frac": 0.0, + "mean_diam_um": 3.0438654815596147, + "area_weighted_mean_diam_um": 7.175146824352199, + "speckle_grain_frac": 0.958695148778511, + "largest_grain_px": 181 + }, + "recovered": { + "config": { + "mis_tol": 5.0, + "min_size": 10, + "fit_max": 1.4, + "erode_iters": 2 + }, + "n_raw_grains": 10885, + "n_grains": 1905, + "retained_frac": 0.9994966596504072, + "cleanup_altered_frac": 0.012423354992221103, + "mean_diam_um": 14.272038360484343, + "area_weighted_mean_diam_um": 19.07078966828683, + "speckle_grain_frac": 0.0, + "largest_grain_px": 545 + } +} \ No newline at end of file diff --git a/example/area2-results/grain-reconstruction/sweep_cleanup.png b/example/area2-results/grain-reconstruction/sweep_cleanup.png new file mode 100644 index 0000000..bd4f7bf Binary files /dev/null and b/example/area2-results/grain-reconstruction/sweep_cleanup.png differ diff --git a/example/area2-results/grain-reconstruction/sweep_min_grain_size.png b/example/area2-results/grain-reconstruction/sweep_min_grain_size.png new file mode 100644 index 0000000..e96b767 Binary files /dev/null and b/example/area2-results/grain-reconstruction/sweep_min_grain_size.png differ diff --git a/example/area2-results/grain-reconstruction/sweep_misorientation_tolerance.png b/example/area2-results/grain-reconstruction/sweep_misorientation_tolerance.png new file mode 100644 index 0000000..3d317fb Binary files /dev/null and b/example/area2-results/grain-reconstruction/sweep_misorientation_tolerance.png differ diff --git a/example/area2-results/grain-reconstruction/sweep_quality_filter.png b/example/area2-results/grain-reconstruction/sweep_quality_filter.png new file mode 100644 index 0000000..c7b15b8 Binary files /dev/null and b/example/area2-results/grain-reconstruction/sweep_quality_filter.png differ diff --git a/example/area2-results/grain-reconstruction/worse_then_better.png b/example/area2-results/grain-reconstruction/worse_then_better.png new file mode 100644 index 0000000..e9c4f98 Binary files /dev/null and b/example/area2-results/grain-reconstruction/worse_then_better.png differ diff --git a/example/area2-results/ipf_map.png b/example/area2-results/ipf_map.png new file mode 100644 index 0000000..470d00d Binary files /dev/null and b/example/area2-results/ipf_map.png differ diff --git a/example/area2-results/master-pattern-comparison/kinematical_vs_dynamical_al.png b/example/area2-results/master-pattern-comparison/kinematical_vs_dynamical_al.png new file mode 100644 index 0000000..e774c7f Binary files /dev/null and b/example/area2-results/master-pattern-comparison/kinematical_vs_dynamical_al.png differ diff --git a/example/area2-results/quality_maps.png b/example/area2-results/quality_maps.png new file mode 100644 index 0000000..9e78d80 Binary files /dev/null and b/example/area2-results/quality_maps.png differ diff --git a/example/area2-results/stats.json b/example/area2-results/stats.json new file mode 100644 index 0000000..900e29d --- /dev/null +++ b/example/area2-results/stats.json @@ -0,0 +1,16 @@ +{ + "n_points": 87416, + "n_indexed": 87163, + "ci_gt_0p1_frac": 0.9950581129312711, + "cm_mean": 0.6478481737963515, + "cm_median": 0.694476306438446, + "fit_mean_deg": 0.5441375877163698, + "fit_median_deg": 0.4931597113609314, + "fit_lt_1p4_frac": 0.9965223757664501, + "index_seconds": 196.02042388916016, + "pc": [ + 0.51783355490249, + 0.33555522935955406, + 0.7526682869865131 + ] +} \ No newline at end of file diff --git a/outputs/edison-closed-loop-ebsd/_task_id.json b/outputs/edison-closed-loop-ebsd/_task_id.json new file mode 100644 index 0000000..b174dbe --- /dev/null +++ b/outputs/edison-closed-loop-ebsd/_task_id.json @@ -0,0 +1,5 @@ +{ + "task_id": "2a641486-1ec5-49ad-b3e8-b2836932d961", + "job": "LITERATURE", + "query": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature." +} \ No newline at end of file diff --git a/outputs/edison-closed-loop-ebsd/answer.md b/outputs/edison-closed-loop-ebsd/answer.md new file mode 100644 index 0000000..2f3c2b3 --- /dev/null +++ b/outputs/edison-closed-loop-ebsd/answer.md @@ -0,0 +1,139 @@ +Question: Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters. + +Specifically map what has been published on: +1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning. +2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set. +3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores — i.e., multi-objective or constrained optimization of EBSD processing. +4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing. +5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size). + +For each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature. + +# Systematic prior-art survey: closed-loop optimization of EBSD processing and indexing + +## Scope and search interpretation + +The survey distinguished four often-confounded categories: **(i)** optimization of pattern-processing or indexing parameters, **(ii)** algorithmic refinement of orientation or pattern-center estimates for a fixed pattern, **(iii)** automated post-indexing map cleanup/segmentation, and **(iv)** automation of acquisition or registration. Only the first category squarely satisfies area 1; automation in the other categories is relevant as neighboring prior art but does not establish a closed-loop EBSD indexing optimizer. + +Searches covered EBSD with Bayesian optimization, Gaussian processes, evolutionary/genetic algorithms, particle swarm optimization, grid/random search, reinforcement learning, Hough parameters, dictionary indexing, reindexing, CI/IQ/MAD/fit/NCC, grain-level objectives, autonomous microscopy, and acquisition feedback. Patent searches returned no relevant EBSD patent in the searched corpus. Absence findings below therefore mean **“not found in this systematic search,” not a legal opinion of novelty or freedom to operate**. + +| Area | Closest published work(s) | What is automated | Objective/criterion | What is explicitly not automated | Gap assessment | +|---|---|---|---|---|---| +| 1. Optimizer-driven tuning of EBSD pattern-processing or indexing parameters | **Closest demonstrated EBSD parameter sweep:** Polonsky et al. 2019 parameterized DREAM.3D reconstruction variables and swept grain-segmentation misorientation tolerance over 0.1–10° in 0.1° increments, generating 100 reconstructions; **nearby but different:** Zhu et al. 2022 is only available here as metadata for “global optimization algorithms” applied to orientation/pattern-center/deformation extraction, so it should be treated as adjacent prior art, not verified from text in this record; **nearby non-tuning automation:** Kaufmann et al. 2020 CNN phase mapping automates phase recognition from patterns rather than tuning EBSD processing settings (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7) | Automated batch execution / parameterization of reconstruction pipelines in BisQue+DREAM.3D; CNN-based direct phase classification from patterns in Kaufmann 2020 (polonsky2019threedimensionalanalysisand pages 1-2, kaufmann2020phasemappingin pages 1-2) | Polonsky used reconstruction-level metrics: fraction of “good” CI-masked voxels retained and total grain count, selecting an elbow/balance point at 2.0°; Kaufmann targets phase-map fidelity qualitatively against known microstructure/EDS, not optimizer search (polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 6-7) | No evidence here of Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, RL, random/grid search over **pattern** processing settings such as background correction, binning, neighbor averaging, Hough band count, rho-theta smoothing, or dictionary settings. Polonsky optimizes segmentation/reconstruction variables after indexing, not indexing settings themselves. Kaufmann does not tune EBSD parameters at all (polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 6-7) | **Gap largely open.** Closest evidence is reconstruction-pipeline parameterization, not closed-loop optimizer-driven tuning of pattern-processing/indexing knobs. | +| 2. Feedback loops that re-index stored raw patterns using CI/IQ/MAD/NCC to choose next parameter set | **Closest proposal:** Shen et al. 2019 proposes combining fast CNN predictions with a confidence measure (quaternion norm) so that high-confidence points get neighborhood dictionary search and low-confidence points get full dictionary search; **related raw-pattern availability, not loop:** Stinville et al. 2022 notes raw .up2 patterns can be opened in EMsoft DI / EMSphInx, enabling reindexing in principle (shen2019convolutionalneuralnetworkbased pages 10-12, shen2019convolutionalneuralnetworkbased pages 1-2, stinville2022multimodaldatasetof pages 9-12, stinville2022multimodaldatasetof pages 12-13) | Shen automates confidence estimation for each location and proposes routing subsequent dictionary matching effort based on confidence; raw pattern archival in Stinville enables offline reindexing workflows (shen2019convolutionalneuralnetworkbased pages 10-12, stinville2022multimodaldatasetof pages 12-13) | Quaternion norm as confidence proxy in Shen; no demonstrated CI/IQ/MAD/NCC-driven next-parameter selection loop in the gathered evidence (shen2019convolutionalneuralnetworkbased pages 10-12) | Shen does **not** demonstrate an executed closed loop that iteratively changes indexing/processing parameters and re-indexes stored raw patterns to maximize CI/IQ/MAD/NCC/NCC-like objectives; it is a proposed hybrid search allocation strategy. Stinville provides data access, not autonomous optimization (shen2019convolutionalneuralnetworkbased pages 10-12, stinville2022multimodaldatasetof pages 12-13) | **Gap open.** Proposal-level hybrid confidence routing exists, but not a demonstrated feedback optimizer that selects the next EBSD parameter set from reindexed raw patterns. | +| 3. Optimization against grain-level or microstructure-level outcomes | **Closest demonstrated:** Polonsky et al. 2019 explicitly optimizes a grain-segmentation parameter against microstructure-level outcomes (grain count and retained good-data fraction); Staněk et al. 2020 compares 2D/3D segmentation sensitivity to misorientation threshold and shows strong threshold dependence, but not autonomous optimization (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8, stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 7-9) | Automated parallel reconstructions and post-processing counts in Polonsky; analytical comparison of threshold effects on segmentation in Staněk (polonsky2019threedimensionalanalysisand pages 5-6, stanek2020comparisonofsegmentation pages 1-2) | Grain count stability / avoidance of grain merging plus retained CI-masked voxel fraction in Polonsky; sensitivity of number of grains and segmentation discrepancies vs threshold in Staněk (polonsky2019threedimensionalanalysisand pages 5-6, stanek2020comparisonofsegmentation pages 7-9) | No demonstrated multi-objective or constrained optimization over broader grain-level cleanup outcomes such as speckle/wild-grain fraction, grain-boundary integrity, fraction of pixels altered by cleanup, or Pareto balancing across multiple EBSD post-processing stages. Polonsky is a one-parameter sweep with manually interpreted elbow criterion (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8) | **Partially occupied, but narrow.** Grain-level optimization exists for segmentation tolerance; broader autonomous multi-objective EBSD post-processing remains a gap. | +| 4. Hybrid Hough-first / dictionary-on-residue and automated escalation criteria | **Closest published work:** Shen et al. 2019 explicitly proposes a hybrid strategy: CNN preliminary prediction plus confidence, then pattern matching only in a neighborhood for high-confidence points and full dictionary search for low-confidence points; this is the clearest automated escalation criterion in the gathered evidence. Sparks et al. 2024 and Winiarski et al. 2021 automate registration/distortion correction workflows but not Hough→dictionary escalation (shen2019convolutionalneuralnetworkbased pages 10-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17) | Automated confidence scoring and proposed routing of dictionary-search scope in Shen; automated image registration/distortion correction in Sparks/Winiarski (shen2019convolutionalneuralnetworkbased pages 10-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17) | Criterion in Shen is confidence from quaternion norm: high confidence → local neighborhood dictionary search; low confidence → full dictionary search (shen2019convolutionalneuralnetworkbased pages 10-12) | No evidence in the gathered text of a **demonstrated** Hough-first then dictionary-on-low-quality-residue loop with automated escalation based on CI/IQ/MAD/NCC thresholds. Shen’s routing is proposed around CNN+dictionary, not Hough+dictionary, and not shown as an executed autonomous residue-reindexing pipeline (shen2019convolutionalneuralnetworkbased pages 10-12) | **Mostly gap.** There is a close proposal for automated escalation, but not a demonstrated Hough-residue-to-dictionary autonomous workflow. | +| 5. Autonomous/self-driving EBSD post-processing or characterization workflows, including links back to SEM acquisition recommendations | **Closest workflow automation:** Sparks et al. 2024 automates sample/image registration steps, pattern-center fitting for every section, and alignment routines; Winiarski et al. 2021 automates SE–EBSD distortion correction and argues more automated procedures are needed; Burnett et al. 2016 demonstrates a fully automated serial-sectioning routine with tracking fiducials, but acquisition settings were experimentally chosen; Kaufmann et al. 2020 automates phase mapping from patterns (sparks20243dreconstructionof pages 9-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17, winiarski2021correctionofartefacts pages 4-6, burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7, kaufmann2020phasemappingin pages 1-2) | Automated registration/warping, per-layer EBSD-to-BSE alignment, CI-threshold masks, pattern-center fitting per section, and automated serial sectioning with fiducial tracking; CNN inference for phase maps (sparks20243dreconstructionof pages 23-24, sparks20243dreconstructionof pages 9-12, burnett2016largevolumeserial pages 4-6, kaufmann2020phasemappingin pages 1-2) | Workflow goals are geometric consistency, low inter-section misorientation, and corrected distortions; Burnett assesses equivalent quality via EBSD pattern quality/band contrast and indexing rate after experimentally selected settings (sparks20243dreconstructionof pages 9-12, winiarski2021correctionofartefacts pages 15-17, burnett2016largevolumeserial pages 6-7) | No evidence here of a human-out-of-the-loop EBSD post-processing system that autonomously tunes post-processing/indexing parameters end-to-end, nor of a loop that feeds post-processing outcomes back to SEM acquisition parameters such as beam current, dwell time, or step size. Burnett explicitly determines acceptable milling current experimentally; Winiarski says more automated procedures still need to be developed (burnett2016largevolumeserial pages 6-7, winiarski2021correctionofartefacts pages 15-17) | **Gap open.** Sub-workflow automation exists, but not a self-driving closed-loop EBSD characterization system that recommends or controls acquisition parameters from post-processing outcomes. | + + +*Table: This table maps the closest published prior art found for the five requested EBSD automation areas, distinguishing demonstrated methods from proposals and from absent closed loops. It highlights that most evidence clusters around workflow automation and grain-segmentation parameter sweeps, not autonomous closed-loop tuning of indexing/processing parameters.* + +## 1. Optimizer-driven tuning of pattern-processing or indexing parameters + +### Closest demonstrated work: Polonsky et al. (2019) + +Polonsky and colleagues integrated DREAM.3D into the cloud-based BisQue platform so that pipeline variables could be parameterized and executions parallelized. Their principal EBSD example varied the **post-indexing grain-segmentation misorientation tolerance** from 0.1° to 10° in 0.1° increments, producing 100 three-dimensional reconstructions while holding minimum-neighbor and minimum-defect-size settings fixed. They evaluated the fraction of CI-masked “good” voxels retained and total reconstructed grain count, then selected 2.0° from convergence/elbow behavior in the derivatives of those metrics. This is genuine automated parameter sweeping with microstructure-level evaluation. However, it operates on already indexed orientation data; the study states that raw EBSD patterns were not collected for that dataset. It therefore does **not** tune background correction, detector binning, neighbor averaging, Hough band count, Hough rho/theta smoothing, or dictionary-search settings, and it cannot re-index patterns under candidate settings. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8) + +### Adjacent work that should not be overread + +Shen et al. developed CNN-based orientation indexing and optimized neural-network architecture/training for speed and prediction accuracy. The trained model indexes patterns directly and uses quaternion-norm deviation as a reliability measure. This replaces or supplements conventional indexing; it does not run Bayesian, evolutionary, grid, or reinforcement-learning search over conventional EBSD processing parameters. It also remains sensitive to experimental geometry and requires detector/setup parameters to be determined beforehand or accommodated through retraining. (shen2019convolutionalneuralnetworkbased pages 1-2, shen2019convolutionalneuralnetworkbased pages 8-10, shen2019convolutionalneuralnetworkbased pages 10-12) + +Kaufmann et al. automated phase classification from individual EBSD patterns with a CNN. The work reduces human decisions in phase mapping and outperforms conventional Hough-plus-user-selected phase/chemistry workflows in difficult examples. The “optimization” described for the conventional comparator was repeated **user adjustment** of EDS reference chemistry against a microstructure already known in advance, not an algorithmic search over EBSD settings. The CNN likewise classifies patterns rather than optimizing background, Hough, binning, or dictionary parameters. (kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7) + +Global/evolutionary optimization has also been published for extracting or refining orientation, pattern center, and deformation state from pattern matching. That is important neighboring art: it establishes optimizer use **inside a per-pattern inverse problem**. It is not equivalent to an outer loop choosing data-processing/indexing hyperparameters from map-quality or grain-quality outcomes. The retrieved full-text evidence did not support treating such work as prior art for the broader closed-loop parameter-selection architecture. + +**Assessment:** No verified publication was found that applies Bayesian optimization, a GP surrogate, genetic/evolutionary search, random/grid search, RL, or a comparable optimizer to the listed **pattern-processing and indexing knobs** and automatically selects the best configuration. Area 1 remains substantially open, aside from adjacent per-pattern refinement and post-indexing segmentation sweeps. + +## 2. Re-indexing stored raw patterns in a metric-driven feedback loop + +Stinville et al. provide a useful enabling precedent: their archived 3D dataset includes approximately 1.71 TB of raw `.up2` Kikuchi patterns, and the paper explicitly states that these can be opened and reprocessed with EMsoft dictionary indexing, EMSphInx spherical-pattern reindexing, or commercial EBSD software. The indexed files also expose IQ and CI. This demonstrates the technical availability of stored-pattern reindexing, but the paper is a dataset publication, not an optimizer: it does not generate candidate parameter sets, score them, or select the next experiment. (stinville2022multimodaldatasetof pages 9-12, stinville2022multimodaldatasetof pages 12-13) + +Shen et al. provide the closest feedback-like decision rule. They propose obtaining a preliminary CNN orientation and confidence at every point, performing only a local dictionary search around high-confidence predictions, and performing a full dictionary search for low-confidence predictions. The confidence proxy is the norm of the predicted quaternion. This is automated computational routing, but it is not a demonstrated iterative loop over processing/indexing parameter sets, and it does not use CI, IQ, MAD/fit, or NCC to propose the next candidate configuration. (shen2019convolutionalneuralnetworkbased pages 10-12) + +Polonsky et al. use CI to define an initial mask of good voxels and assess how many remain after cleanup and segmentation. CI is therefore part of a reconstruction-level quality assessment, but it is not fed back to re-index raw patterns; indeed, raw patterns were unavailable in that example. (polonsky2019threedimensionalanalysisand pages 5-6) + +**Assessment:** No demonstrated publication was found with the complete loop **candidate parameters → re-index stored raw patterns → compute CI/IQ/MAD/fit/NCC objective → choose the next parameter set → repeat**. Area 2 remains an unclaimed gap in the searched literature. + +## 3. Grain- or microstructure-level objectives + +This area has the strongest direct prior art, but it is narrow. + +Polonsky et al. explicitly recognized that segmentation tolerance changes grain consolidation and retained data in opposite directions. Their two high-level metrics were total grain count and the fraction of initially CI-qualified voxels retained through reconstruction. They chose the lowest tolerance at which the changes in both quantities became approximately constant, analogous to an elbow criterion. This is a microstructure-level, effectively two-metric parameter-selection procedure rather than optimization of a per-pattern score. (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8) + +Staněk et al. independently show why such objectives matter. They analyzed the dependence of 2D and 3D grain segmentation on misorientation threshold, including mean grain counts and the formation of erroneously merged giant grains. For one noisy experimental dataset, changing noise removal and excluding grains below eight voxels changed the detected population from more than one million grains to 74,579. This is a systematic sensitivity study and supplies possible grain-level observables, but it does not implement an autonomous optimizer or select cleanup parameters through a formal multi-objective rule. (stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 7-9) + +Neither publication demonstrates optimization across a broad EBSD-processing chain using objectives such as wild-grain/speckle fraction, boundary continuity or integrity, grain-count stability over threshold perturbations, fraction of pixels modified by cleanup, or a constraint on topology preservation. Polonsky varies essentially one segmentation parameter, keeps other settings fixed, and identifies a balance/elbow rather than computing a Pareto front or solving a constrained optimization problem. (polonsky2019threedimensionalanalysisand pages 5-6) + +**Assessment:** Area 3 is **partially occupied**. A broad claim to “optimizing EBSD processing using grain-level outcomes” would encounter Polonsky et al. A narrower architecture that jointly tunes indexing and cleanup parameters using several topology- and intervention-aware grain objectives still appears open in the searched literature. + +## 4. Hybrid Hough-first/dictionary-on-residue indexing + +The closest automated escalation rule is Shen et al., but its first stage is a CNN rather than Hough indexing. Their proposed rule is explicit: high-confidence CNN results receive dictionary matching only near the preliminary orientation; low-confidence results receive full dictionary search. Thus, the publication anticipates confidence-triggered allocation of expensive pattern matching. It does not demonstrate a Hough-first pipeline, does not define residue using CI/IQ/MAD/NCC from Hough indexing, and does not report an executed end-to-end selective reindexing experiment. (shen2019convolutionalneuralnetworkbased pages 10-12) + +The same paper compares CNN and Hough indexing at different exposure times. CNN indexing produced a larger fraction of points below a 10° misorientation-error cutoff at high noise, whereas Hough was more accurate among the smaller subset it indexed correctly. This comparison motivates selective hybridization, but the reported hybrid suggestion remains CNN-to-dictionary rather than Hough-to-dictionary. (shen2019convolutionalneuralnetworkbased pages 8-10, shen2019convolutionalneuralnetworkbased pages 10-12) + +Kaufmann et al. likewise replace conventional Hough phase assignment with CNN classification; they do not retain Hough as a first pass and selectively escalate failed pixels. (kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7) + +**Assessment:** Confidence-triggered hybrid search is neighboring prior art, so a very broad claim to “use confidence to decide whether to run dictionary matching” is not open. However, no verified demonstration was found of **Hough first → automatically identify low-quality residue → dictionary/pattern-match only that residue**, particularly using an explicit CI/IQ/MAD/NCC or grain-context escalation criterion. Area 4 therefore remains mostly open in the requested Hough-specific form. + +## 5. Autonomous EBSD workflows and acquisition feedback + +Several publications automate substantial sub-workflows: + +* **BisQue–DREAM.3D:** cloud execution, provenance, parallel pipeline runs, and range parameterization for 3D reconstruction. Parameters and ranges are supplied rather than autonomously discovered, and the demonstrated optimization begins after indexing. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6) +* **Automated serial sectioning:** Burnett et al. describe a fully automated PFIB serial-section routine using automatically recognized tracking fiducials and repeated milling/imaging positions. Yet acceptable milling current was determined experimentally, and the authors state that systematic mapping of voltage/current space remained necessary. EBSD settings—including beam energy/current, dwell, binning, band count, and Hough settings—were fixed by the investigators rather than adapted from post-processing results. (burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7) +* **Automated registration and distortion correction:** Sparks et al. use automated sample alignment, pattern-center fitting for each section, CI thresholding, and optimizer-based multimodal registration. The optimizer maximizes an image-registration similarity metric and transforms maps; it does not alter EBSD indexing settings or SEM acquisition conditions. (sparks20243dreconstructionof pages 9-12, sparks20243dreconstructionof pages 23-24) +* **SE–EBSD map correction:** Winiarski et al. automate aspects of co-registration and long-/short-range distortion correction using trainable filters and moving least squares. The paper explicitly concludes that more automated procedures and more robust accuracy measures still need development. It also discusses faster or smart acquisition as a prospect, not as a closed post-processing-to-acquisition controller. (winiarski2021correctionofartefacts pages 4-6, winiarski2021correctionofartefacts pages 15-17) +* **CNN indexing/phase mapping:** Shen and Kaufmann automate inference from patterns, but training data, instrument calibration, phase standards, and/or model selection remain external; neither closes the loop to microscope settings. (shen2019convolutionalneuralnetworkbased pages 1-2, shen2019convolutionalneuralnetworkbased pages 8-10, kaufmann2020phasemappingin pages 1-2) + +No retrieved work used post-processing outcomes to recommend or automatically change **beam current, pattern dwell/exposure, detector binning, or spatial step size** for a subsequent SEM/EBSD acquisition. Burnett et al. come closest to connecting acquisition conditions with indexing rate and band contrast, but the comparison is an investigator-designed experiment, not a controller: selected ion currents are evaluated under fixed EBSD conditions and an acceptable current is chosen experimentally. (burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7) + +**Assessment:** Automated components and scripted pipelines are well established, so “automated EBSD workflow” alone is not a gap. A genuinely autonomous, human-out-of-the-loop EBSD characterization system that interprets indexed/map/grain quality and then recommends or controls SEM acquisition parameters remains open in the searched literature. + +## Consolidated gap assessment + +1. **Optimizer-driven tuning of EBSD pattern-processing/indexing settings — open.** Neighboring art covers per-pattern inverse optimization, ML indexing, and post-indexing segmentation sweeps, but not outer-loop optimization of background, binning, averaging, Hough, or dictionary hyperparameters. +2. **Metric-driven iterative reindexing of stored raw patterns — open.** Raw-pattern reindexing is enabled and confidence-directed search is proposed, but the complete iterative objective loop was not found. +3. **Grain-level optimization — partially claimed.** Polonsky et al. directly cover segmentation-tolerance selection using retained-good-data fraction and grain count. Broad multi-objective, constrained, topology-aware optimization across indexing plus cleanup remains open. +4. **Automated hybrid escalation — partially anticipated, Hough-specific implementation open.** Shen et al. anticipate confidence-triggered local versus full dictionary search after CNN indexing. A demonstrated Hough-first, dictionary-on-residue workflow with formal escalation criteria was not found. +5. **Autonomous workflow with feedback to acquisition — open at the system level.** Serial sectioning, registration, reconstruction, correction, and ML inference are individually automated, but no end-to-end post-processing-to-SEM acquisition feedback system was found. + +The most defensible unexplored combination is therefore: **repeated reindexing of archived raw patterns under optimizer-selected processing/indexing settings; joint pattern-, pixel-, and grain-level objectives with explicit cleanup/topology constraints; confidence- and grain-context-based escalation from Hough to dictionary matching; and translation of the optimized result into acquisition recommendations or microscope control.** None of the verified publications combines those elements into a closed, autonomous EBSD loop. + +References + +1. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +2. (polonsky2019threedimensionalanalysisand pages 5-6): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +3. (kaufmann2020phasemappingin pages 1-2): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal. + +4. (kaufmann2020phasemappingin pages 6-7): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal. + +5. (shen2019convolutionalneuralnetworkbased pages 10-12): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal. + +6. (shen2019convolutionalneuralnetworkbased pages 1-2): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal. + +7. (stinville2022multimodaldatasetof pages 9-12): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal. + +8. (stinville2022multimodaldatasetof pages 12-13): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal. + +9. (polonsky2019threedimensionalanalysisand pages 6-8): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +10. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. + +11. (stanek2020comparisonofsegmentation pages 7-9): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. + +12. (sparks20243dreconstructionof pages 23-24): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal. + +13. (winiarski2021correctionofartefacts pages 15-17): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal. + +14. (sparks20243dreconstructionof pages 9-12): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal. + +15. (winiarski2021correctionofartefacts pages 4-6): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal. + +16. (burnett2016largevolumeserial pages 4-6): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal. + +17. (burnett2016largevolumeserial pages 6-7): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal. + +18. (shen2019convolutionalneuralnetworkbased pages 8-10): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal. \ No newline at end of file diff --git a/outputs/edison-closed-loop-ebsd/references.txt b/outputs/edison-closed-loop-ebsd/references.txt new file mode 100644 index 0000000..70977d5 --- /dev/null +++ b/outputs/edison-closed-loop-ebsd/references.txt @@ -0,0 +1,35 @@ +1. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +2. (polonsky2019threedimensionalanalysisand pages 5-6): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +3. (kaufmann2020phasemappingin pages 1-2): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal. + +4. (kaufmann2020phasemappingin pages 6-7): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal. + +5. (shen2019convolutionalneuralnetworkbased pages 10-12): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal. + +6. (shen2019convolutionalneuralnetworkbased pages 1-2): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal. + +7. (stinville2022multimodaldatasetof pages 9-12): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal. + +8. (stinville2022multimodaldatasetof pages 12-13): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal. + +9. (polonsky2019threedimensionalanalysisand pages 6-8): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +10. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. + +11. (stanek2020comparisonofsegmentation pages 7-9): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. + +12. (sparks20243dreconstructionof pages 23-24): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal. + +13. (winiarski2021correctionofartefacts pages 15-17): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal. + +14. (sparks20243dreconstructionof pages 9-12): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal. + +15. (winiarski2021correctionofartefacts pages 4-6): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal. + +16. (burnett2016largevolumeserial pages 4-6): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal. + +17. (burnett2016largevolumeserial pages 6-7): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal. + +18. (shen2019convolutionalneuralnetworkbased pages 8-10): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal. \ No newline at end of file diff --git a/outputs/edison-closed-loop-ebsd/task_full.json b/outputs/edison-closed-loop-ebsd/task_full.json new file mode 100644 index 0000000..b1ee5f1 --- /dev/null +++ b/outputs/edison-closed-loop-ebsd/task_full.json @@ -0,0 +1,5983 @@ +{ + "status": "success", + "query": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "user": null, + "created_at": "2026-08-11T22:00:10.593150Z", + "job_name": "job-futurehouse-paperqa3", + "share_status": "private", + "permitted_accessors": { + "users": [], + "organizations": [] + }, + "build_owner": "FutureHouse", + "environment_name": null, + "agent_name": null, + "task_id": "2a641486-1ec5-49ad-b3e8-b2836932d961", + "project_id": null, + "agent_state": [ + { + "id": "828d8ade-4946-483a-8ece-4a37c126e034", + "trajectory_id": "2a641486-1ec5-49ad-b3e8-b2836932d961", + "trajectory_timestep": 12, + "step": "after_transition", + "state": { + "transition": { + "timestep": 12, + "agent_state": { + "tools": [ + { + "type": "function", + "info": { + "name": "paper_search", + "description": "Search for papers in Google Scholar for you to look at.\n\nCan be called repeatedly with identical parameters to retrieve more results (max 2 repeats).\nA short summary of the retrieved papers will be returned, but to examine the full content of each paper,\nyou must use the gather_evidence tool.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A Google Scholar search query, which can be a specific phrase, complete sentence,\nor general keywords, e.g. 'machine learning for immunology'. Also can be\ngiven search operators like 'author:'.", + "title": "Query", + "type": "string" + }, + "min_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for minimum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Min Year" + }, + "max_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for maximum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Max Year" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "gather_evidence", + "description": "Given a specific question, gather evidence from full-text of documents that have been retrieved by\nprevious tool calls.\n\nHere is how the tool works: the provided question is posed to all texts retrieved thus far\nby other tools. An LLM sees each text independently and is prompted to find information relevant\nto the question.\n\nThat is, we sample evidence `e \\sim LLM(prompt, text, q)` for each text. We only return pieces of\nevidence `e` that are sufficiently relevant to the question; if the text is irrelevant, then `e`\nis not included in the final response of this tool. Note that this LLM does not see the conversation\nhistory thus far, only the question and each text. Note that it only sees text, no images or other\nmedia from the contents (though tables and formulae may be parseable in some cases).\n\nThink of this tool as an efficient way to deeply scan all retrieved documents for evidence. It\nis complementary to the `read_text` tool, which can be used to directly inspect a subset of the\nretrieved documents.\n\nA valuable time to invoke this tool is after another tool finds potentially relevant documents.\nFeel free to invoke this tool in parallel with other tools, but do not call this tool in parallel with itself.\nOnly invoke this tool when the document count is above zero, or this tool will be useless.\n\nIf you call this tool 2+ times and repeatedly get 'No relevant evidence found',\nconsider whether the answer might be in visual content (figures/tables) rather than text,\nand use appropriate tools if available to access that content.\n\nOnce you receive evidence from this tool:\n\n1. Read Carefully: Don't skim. The evidence may contain multiple similar terms\n or values\u2014make sure you identify the RIGHT one.\n\n2. For Comparative Questions: If the question asks \"which X has greatest Y\",\n check that the evidence includes information about ALL relevant X options, not\n just one or two. If you only see partial information, gather more evidence.\n\n3. Check for Noise: Verify that numbers or terms aren't just metadata\n (page numbers, reference IDs, dates). Look for the value in meaningful\n scientific context.\n\n4. Context Matters: A term appearing 71 times doesn't mean it's always the answer.\n Verify which occurrence actually answers the question asked.\n\nIf the evidence doesn't clearly answer the question, gather more evidence before\nconcluding. Don't guess based on partial information.", + "parameters": { + "type": "object", + "properties": { + "question": { + "description": "Specific question to gather evidence for.", + "title": "Question", + "type": "string" + }, + "text_name_focus_list": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "if there are known text names that are relevant to the question,\nyou can specify them here. This will ensure that the tool will look at these text names\nfirst, and then look at the rest of the evidence. If left as `None`, an embedding\nsearch will be performed on underlying content to find most relevant texts.", + "title": "Text Name Focus List" + } + }, + "required": [ + "question" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "open_targets_search", + "description": "Search for disease-target associations from OpenTargets database.\n\nAdds new disease-target association data to the state and returns top search results.", + "parameters": { + "type": "object", + "properties": { + "disease_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of disease names to search for.\nUsage:\ndisease_names = [\"Alzheimer's\"] ->\n Returns top targets for Alzheimer's\ndisease_names = [\"Alzheimer's\", \"Diabetes\"], target_names = [\"PSEN1\"] ->\n Filters for only PSEN1-Alzheimer's/Diabetes evidence\ndisease_names = [\"Alzheimer's\"], drug_names = [\"Aducanumab\"] ->\n Filters for Alzheimer's disease-target pairs treated by Aducanumab", + "title": "Disease Names" + }, + "target_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of target names to search for.\nUsage:\ntarget_names = [\"BACE1\"] ->\n Returns top diseases associated with BACE1\ntarget_names = [\"BACE1\", \"APP\"], disease_names = [\"Alzheimer's\"] ->\n Filters for only these target-Alzheimer's pairs\ntarget_names = [\"BACE1\"], drug_names = [\"Aducanumab\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Target Names" + }, + "drug_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of drug names to generate or filter results.\nUsage:\ndrug_names = [\"Aducanumab\"] ->\n Returns all disease-target pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], disease_names = [\"Alzheimer's\"] ->\n Filters for Alzheimer's pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], target_names = [\"BACE1\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Drug Names" + } + }, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "collect_cited_papers_in_evidence", + "description": "Collect papers by traversing the citations of relevant papers to increase the paper count.\n\nThis tool has no effect if called when paper count or relevant papers are zero.\nThis tool will find papers that are likely to lead to more relevant evidence.\nThe papers found have a higher chance of relevance because this tool starts from relevant evidence.\nThis tool's usefulness increases with the amount of gathered evidence as there's more citations to consider.", + "parameters": { + "type": "object", + "properties": {}, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "view_images", + "description": "Retrieve images from a document.\n\nUse this tool when you need to see figures, tables, or other visual content.\nThe tool works by asking a weaker LLM agent to scan through the document, looking for\nvisual content matching the query you provide. When the agent finds matches, it will\ncrop the page to the relevant region. It can be tasked to return multiple images,\nif desired. Note that the weaker LLM may be worse at *understanding* images than\nyou, so double-check its conclusions about the returned images. It is also stateless,\ni.e. it will not remember what you have asked it to find in the past.\n\nUse this tool carefully. It is difficult to retrieve images, and the images\nthemselves will fill up your context window quickly. Here are some guidelines:\n1. ALWAYS use this tool if the question explicitly mentions 'Table X' or 'Figure Y'.\n These questions require visual content and cannot be answered with text alone.\n2. Use this tool as a fallback when gather_evidence fails repeatedly (2+ times\n returning 'No relevant evidence'). This suggests the answer may be in visual content.\n3. When gather_evidence returns text mentioning 'see Figure X' or 'shown in Table Y',\n you should call this tool to examine that specific visual reference.\n4. Only call this tool after directly inspecting relevant text with read_text\n to confirm the document likely contains the visual you need.\n5. Don't use this tool in a broad-spectrum way (don't call on many documents\n just to see what might be relevant).\n6. Never call this tool in parallel with itself.\n7. Users like visual evidence, so call this tool at least once before answering so you\n can provide some image citations in the answer. Add this to your plan so you do not forget.\n For images to be presented to the user, they must be cited in the final answer after retrieval\n by this tool. You may ignore this instruction if the request is for the question is asking\n for a specific piece of information and is not an open-ended research question.\n8. NEVER cite full-page images in your final answer, unless they are full-page panels figures or\n tables. Cite relevant cropped regions if available, otherwise consider calling this tool again\n to find a smaller region that answers the query.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "Query to search for visual content in the document. If you know the specific\nfigure/table/formula you are looking for, you can specify it here (e.g. \"find me\nFigure 3\" or \"extract the table on page 10\"). You can also request based on content\n(e.g. \"find me all volcano plots of gene X\" or \"I need the formula for the growth rate\").\nYou can even include explicit instructions for the LLM agent to follow (if helpful, not\nnecessarily required). The more specific and clear your query is here, the higher the\nchance of success.", + "title": "Query", + "type": "string" + }, + "text_name": { + "description": "Name of a text chunk (e.g., 'smith2023example pages 1-5').", + "title": "Text Name", + "type": "string" + } + }, + "required": [ + "query", + "text_name" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "create_artifact", + "description": "Create an artifact that can be used in the final answer, based on information\nyou have collected so far.\n\nHow it works: the full message history thus far is passed to another LLM,\nalong with the question and your description. The LLM is tasked to create\nan artifact that matches your specifications, including a title, caption,\nand markdown content. You drive the process by specifying the type and\ndescribing what you want.\n\nOnly use this tool *after* you have gathered sufficient evidence or\ninformation to create a new artifact. Do not call this tool in parallel\nwith tools that are searching for information or evidence that would be\nnecessary to create this artifact.\n\nIt is very helpful after calls to several search tools,\nafter calls to the `gather_evidence` tool, but it is not required.\nHowever, it should be called before the `answer` tool at\nleast one time, but more may be given if specifically helpful for\nthe task, so that the `answer` tool can use the artifacts for the answer.\n\nAn artifact can be any string representable fact or list of facts,\nformatted using markdown syntax. The artifact will be generated as one of\nthe following types:\n- table: Markdown tables with rows and columns (e.g., findings, methods, molecules)\n- code_block: Code snippets in fenced code blocks\n- blockquote: Quoted text or important facts (using > prefix)\n- image: AI-generated image (chart, diagram, or visualization)\n\nEach artifact consists of:\n- A short title (3-8 words) that names what the artifact contains\n- A caption (1-2 sentences) describing what the artifact shows and why it's useful\n- Markdown content in a single element type (validated automatically with retries if needed)\n For image type, content is a markdown image embed ``![title](data:image/png;base64,...)``,\n embedding the PNG directly as a base64 data URL.", + "parameters": { + "type": "object", + "properties": { + "artifact_type": { + "description": "Artifact type to create. Must be one of: table,\ncode_block, blockquote, image.", + "title": "Artifact Type", + "type": "string" + }, + "description": { + "description": "Description of what you want in the artifact. Be specific\nabout what information should be included, how it should be organized,\nand what columns/sections/content you want. This guides the summary LLM.\nFor image type, describe the visual you want generated in detail.", + "title": "Description", + "type": "string" + } + }, + "required": [ + "artifact_type", + "description" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "update_plan", + "description": "Update the plan for a specific or new objective using a JSON string to create or\nreplace the existing objective.\n\nThe updated_json should be a JSON string representing a single plan objective object.\n\nYou may use this tool to update an objective, create a new one, or change\nstrategy if the current plan is not working. Only change strategy significantly\nas a last resort, after trying very hard to make the current plan work.\n\nOnly mark an objective as completed if you are absolutely sure that it has been fully\nmet. If it is only partially met, you may add additional objectives to the plan for\nwhat is missing.", + "parameters": { + "type": "object", + "properties": { + "plan_objective_id": { + "description": "The id of the objective to update.", + "title": "Plan Objective Id", + "type": "integer" + }, + "updated_plan_objective_json": { + "description": "A JSON string representing the updated objective.\nIt should be a dictionary with the with the following keys:\n- \"id\": (int) an integer id matching plan_objective_id, if not found, a new objective is created.\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Updated Plan Objective Json", + "type": "string" + } + }, + "required": [ + "plan_objective_id", + "updated_plan_objective_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "clinical_trials_search", + "description": "Search for clinical trials, with support for repeated calls and concurrent execution.\n\nWill add new clinical trials to the state, and return metadata about the number of trials found.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "The search query string. Supports complex boolean expressions, field-specific\nsearches, and query modifiers through operators. All configuration is done through\noperators in the query string.\nQuery Syntax:\n Basic Search:\n Simple text automatically uses default EXPANSION[Relaxation] and COVERAGE[Contains]\n >>> \"heart attack\"\n\n Modified Search:\n Use operators to modify search behavior:\n >>> 'EXPANSION[None]COVERAGE[FullMatch]\"exact phrase\"'\n >>> 'EXPANSION[Concept]heart attack'\n\n Field Search:\n Specify fields using AREA operator:\n >>> 'AREA[InterventionName]aspirin'\n >>> 'AREA[Phase]PHASE3'\n\n Location Search:\n Use SEARCH operator for compound location queries:\n >>> 'cancer AND SEARCH[Location](AREA[LocationCity]Boston AND AREA[LocationState]Massachusetts)'\n\n Complex Boolean:\n Combine terms with AND, OR, NOT and parentheses:\n >>> '(cancer OR tumor) AND NOT (EXPANSION[None]pediatric OR AREA[StdAge]CHILD)'\n\n Date Ranges:\n Use RANGE to specify date ranges with formats like \"yyyy-MM\" or \"yyyy-MM-dd\".\n Note that MIN and MAX can be used for open-ended ranges:\n >>> AREA[ResultsFirstPostDate]RANGE[2015-01-01, MAX]\n\nOperators:\n EXPANSION[type]: Controls term expansion\n - None: Exact match only, case and accent sensitive\n - Term: Includes lexical variants (plurals, spellings)\n - Concept: Includes UMLS synonyms\n - Relaxation: Relaxes adjacency requirements (default)\n - Lossy: Allows missing partial terms\n\n COVERAGE[type]: Controls text matching\n - FullMatch: Must match entire field\n - StartsWith: Must match beginning of field\n - EndsWith: Must match end of field\n - Contains: Must match part of field (default)\n\n AREA[field]: Specifies field to search\n - See Field Reference for available fields\n\n SEARCH[type]: Groups field searches\n - Location: Groups location-related fields\n - Study: Groups study-related fields\n\nUsage Notes:\n - All search expressions are implicitly OR expressions\n - Operator precedence (highest to lowest): terms/source operators, NOT/context operators, AND, OR\n - Use quotes for exact phrase matching: \"heart attack\"\n - Use parentheses for grouping: (heart OR cardiac) AND attack\n - Use backslash to escape operators: \\AND\n - Default expansion is EXPANSION[Relaxation]\n - Default coverage is COVERAGE[Contains]\n\nField Reference:\n High Priority Fields (weight >= 0.8):\n - NCTId (1.0): Trial identifier\n - Acronym (1.0): Study acronym\n - BriefTitle (0.89): Short title\n - OfficialTitle (0.85): Full official title\n - Condition (0.81): Medical condition\n - InterventionName (0.8): Primary intervention name\n - OverallStatus: Trial status\n\n Medium Priority Fields (0.5-0.79):\n - InterventionOtherName (0.75): Alternative intervention names\n - Phase (0.65): Trial phase\n - StdAge (0.65): Standard age groups\n - Keyword (0.6): Study keywords\n - BriefSummary (0.6): Short description\n - SecondaryOutcomeMeasure (0.5): Secondary outcomes\n\n Low Priority Fields (< 0.5):\n - DesignPrimaryPurpose (0.3): Primary purpose of study\n - StudyType (0.3)\n - Various descriptive, location, and administrative fields\n\nSupported Enums:\n Phase:\n - EARLY_PHASE1: Early Phase 1\n - PHASE1: Phase 1\n - PHASE2: Phase 2\n - PHASE3: Phase 3\n - PHASE4: Phase 4\n - NA: Not Applicable\n\n StandardAge:\n - CHILD: Child\n - ADULT: Adult\n - OLDER_ADULT: Older Adult\n\n Status:\n - RECRUITING: Currently recruiting participants\n - ACTIVE_NOT_RECRUITING: Active but not recruiting\n - COMPLETED: Study completed\n - ENROLLING_BY_INVITATION: Enrolling by invitation only\n - NOT_YET_RECRUITING: Not yet recruiting\n - SUSPENDED: Study suspended\n - TERMINATED: Study terminated\n - WITHDRAWN: Study withdrawn\n - AVAILABLE: Available\n - NO_LONGER_AVAILABLE: No longer available\n - TEMPORARILY_NOT_AVAILABLE: Temporarily not available\n - APPROVED_FOR_MARKETING: Approved for marketing\n - WITHHELD: Withheld\n - UNKNOWN: Unknown status\n\n StudyType:\n - INTERVENTIONAL: Interventional studies\n - OBSERVATIONAL: Observational studies\n - EXPANDED_ACCESS: Expanded access studies\n\n PrimaryPurpose:\n - TREATMENT: Treatment\n - PREVENTION: Prevention\n - DIAGNOSTIC: Diagnostic\n - ECT: Educational/Counseling/Training\n - SUPPORTIVE_CARE: Supportive Care\n - SCREENING: Screening\n - HEALTH_SERVICES_RESEARCH: Health Services Research\n - BASIC_SCIENCE: Basic Science\n - DEVICE_FEASIBILITY: Device Feasibility\n - OTHER: Other\n\n InterventionType:\n - BEHAVIORAL: Behavioral interventions\n - BIOLOGICAL: Biological interventions\n - COMBINATION_PRODUCT: Combination product interventions\n - DEVICE: Device interventions\n - DIAGNOSTIC_TEST: Diagnostic test interventions\n - DIETARY_SUPPLEMENT: Dietary supplement interventions\n - DRUG: Drug interventions\n - GENETIC: Genetic interventions\n - PROCEDURE: Procedure interventions\n - RADIATION: Radiation interventions\n - OTHER: Other interventions\n\n DesignAllocation:\n - RANDOMIZED: Randomized allocation\n - NON_RANDOMIZED: Non-randomized allocation\n - NA: Not applicable\n\n InterventionalAssignment:\n - SINGLE_GROUP: Single group assignment\n - PARALLEL: Parallel assignment\n - CROSSOVER: Crossover assignment\n - FACTORIAL: Factorial assignment\n - SEQUENTIAL: Sequential assignment\n\n ObservationalModel:\n - COHORT: Cohort\n - CASE_CONTROL: Case-Control\n - CASE_ONLY: Case-Only\n - CASE_CROSSOVER: Case-Crossover\n - ECOLOGIC_OR_COMMUNITY: Ecologic or Community\n - FAMILY_BASED: Family-Based\n - DEFINED_POPULATION: Defined Population\n - NATURAL_HISTORY: Natural History\n - OTHER: Other\n\n DesignMasking:\n - NONE: None (Open Label)\n - SINGLE: Single\n - DOUBLE: Double\n - TRIPLE: Triple\n - QUADRUPLE: Quadruple\n\n WhoMasked:\n - PARTICIPANT: Participant\n - CARE_PROVIDER: Care Provider\n - INVESTIGATOR: Investigator\n - OUTCOMES_ASSESSOR: Outcomes Assessor", + "title": "Query", + "type": "string" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "create_plan", + "description": "Create a plan for to construct a thorough and accurate answer.\n\nDecompose the input task into a set of sub-objectives that need to\nbe met in order to form the answer. Subsequently, when working on solving\nthe query, you can focus on these objectives one by one. Note that the\n`update_plan` tool allows you to change the objectives if needed.\n\nIn addition to the objectives you deem necessary, add one final objective\nto call the `answer` tool with your final answer, so that you do not\nforget to submit your answer.", + "parameters": { + "type": "object", + "properties": { + "plan_json": { + "description": "A JSON string representing the plan. It should be a dictionary\nwith top-level key \"objectives\", whose value is a list of objectives.\nEach objective should be a dictionary with the following keys:\n- \"id\": (int) an integer id for the plan objective\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Plan Json", + "type": "string" + } + }, + "required": [ + "plan_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "read_text", + "description": "Given a list of text names, read their contents.\n\nThis tool is valuable to read the full contents of specific parts of documents,\ncomplementing the `gather_evidence` tool's indirect approach.\nThis can be an especially useful tool if `gather_evidence` does not find evidence,\nbut you think it is very likely that specific documents are still relevant to the question.", + "parameters": { + "type": "object", + "properties": { + "text_name_list": { + "description": "A list of up to 10 text names to read.\nThese text names can be any valid text names retrieved by other tools.", + "items": { + "type": "string" + }, + "title": "Text Name List", + "type": "array" + } + }, + "required": [ + "text_name_list" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "patent_search", + "description": "Search for patents to increase the patent count.\n\nRepeat previous calls with the same query to continue a search.\nOnly repeat a maximum of three times.\nThis tool can be called concurrently.\nThis tool introduces novel patents, so invoke it when just beginning\nor when unsatisfied with the current evidence.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A search query for patents (using Google Patents). Never include temporal\nfilters here (e.g. specific years or dates); instead use the `after` and `before`\narguments. Never include author filters here; instead use the `inventor` argument.\nSupports the following syntax, but they can be brittle and are not preferred; try\nsimple queries where possible:\n- Boolean operators: AND (default), OR, and - (NOT)\n- Exact phrases: \"quoted text\"\n- Proximity: NEAR/x, ADJ/x (ordered), WITH (20 words), SAME (200 words)\n- Field search: TI=(title), AB=(abstract), CL=(claims)\n- CPC codes: cpc:A01B or CPC=B60R22/low (includes children)\n- Wildcards: ? (0-1 char), * (0+ chars), # (exactly 1 char)\n- Chemistry: exact molecule names, SSS=molecule (substructure),\n ~molecule (similar), SMILES strings, SMARTS=pattern, InChI keys,\n CL=~molecule (chemistry in claims)", + "title": "Query", + "type": "string" + }, + "after": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Minimum patent date. Format: YYYYMMDD.\nExample: '20150101' for patents after Jan 1, 2015.", + "title": "After" + }, + "before": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Maximum patent date. Format: YYYYMMDD.\nExample: '20200101' for patents before Jan 1, 2020.", + "title": "Before" + }, + "date_type": { + "default": "publication", + "description": "Type of date that after and before refer to. Must be either 'publication'\nor 'filing'. Only necessary to set if after and/or before are set.", + "title": "Date Type", + "type": "string" + }, + "inventor": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Filter by inventor name(s). Comma-separated for multiple.", + "title": "Inventor" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "answer", + "description": "Submit your final answer to the user. You MUST call this tool on your final answer.\nIf you mark all objectives as completed but have not yet called this tool, then\nthe answer will not be submitted. Note that this tool should only be called once, as it\nimmediately terminates the run and submits the answer.\n\nWrite an answer based on the evidence from gather_evidence results in message history.\nIf the context provides insufficient information, provide a partial answer and explain\nwhy it is partial (e.g. the question is currently an open question in the field,\nyou were unable to obtain certain papers, etc.).\n\nFor each part of your answer, indicate which sources most support it via citation IDs at\nthe end of sentences, like (pqac-0000ABCD).\nYou are only allowed to cite available context IDs, indicated in previous messages by \"Context IDs that may be cited:\".\n## Valid citation examples, only use comma/space delimited parentheticals:\n\n- (pqac-0000ABCD, pqac-0000EFGH)\n- (pqac-0000IJKL)\n\n## Invalid citation examples:\n\n- (pqac-0000ABCD and pqac-0000EFGH)\n- (pqac-0000ABCD;pqac-0000EFGH)\n- (pqac-0000ABCD-pqac-0000EFGH)\n- pqac-0000ABCD and pqac-0000EFGH\n- Example's work pqac-0000ABCD\n- (pages pqac-0000ABCD)\n- ![description](pqac-0000ABCD)\n- ()\n\nDo not concatenate citation ids, just use them as-is.\n\nWhen writing your answer, embed artifacts if relevant artifacts have been previously\ncreated by the `create_artifact` tool. To embed an artifact, simply include this in your\nanswer, on its own line and without any additional formatting (i.e. no backticks, code fences,\nquotes, additional tags, etc.).\nAlways provide at least some introductory sentence or paragraph before referring to an artifact,\nso that the artifact is properly contextualized.\nHere is an example reference to an artifact with ID \"artifact-XY\":\n\n\nValid artifact IDs are indicated in previous messages by 'Artifact IDs that may be injected into the answer:'.\n\n\nDouble-check your answer:\n\n1. Evidence Match: Re-read the specific evidence snippets you're citing. Does your\n answer actually match what they say? Don't rely on memory from 20+ messages ago.\n\n2. For Comparative Questions (\"which X has greatest/maximum/highest Y\"):\n - Did you examine ALL options mentioned in the evidence?\n - Did you explicitly compare their values?\n - Example: If evidence mentions \"Pancreatic: 0.92, Liver: 0.85, Lung: 0.78\",\n verify you compared all three, not just two.\n\n3. For Specific Value Questions (\"How many\", \"What is the\", specific terms):\n - Does the exact value you're stating appear in the evidence?\n - Are you extracting from the right context (not noise like page numbers)?\n\n4. Sanity Check: Read your answer. Does it directly address what was asked?\n\nDon't use Markdown image syntax like ![...](...) to reference media in your answer,\nbecause media get automatically linked by the system based on your answer's citations.\n\nWrite in the style of a scientific article, with concise sentences and\ncoherent paragraphs. This answer will be used directly,\nso do not add any extraneous information.", + "parameters": { + "type": "object", + "properties": { + "answer": { + "description": "valid answer, cited with citation IDs.", + "title": "Answer", + "type": "string" + } + }, + "required": [ + "answer" + ] + } + } + } + ], + "messages": [ + { + "role": "user", + "content": "Use the tools to answer the question: Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.\n\nStart by making a plan, and update the plan as you go. Your answer will be returned to the user when all plan objectives are completed. Be thorough and ensure your evidence matches the user's question.", + "info": null + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_TMwBuCDwZNGWGVmRuRrINX4G", + "type": "function", + "function": { + "arguments": "{\"plan_json\": \"{\\\"objectives\\\":[{\\\"id\\\":1,\\\"objective\\\":\\\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\\\",\\\"rationale\\\":\\\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":2,\\\"objective\\\":\\\"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\\\",\\\"rationale\\\":\\\"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":3,\\\"objective\\\":\\\"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\\\",\\\"rationale\\\":\\\"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":4,\\\"objective\\\":\\\"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\\\",\\\"rationale\\\":\\\"Area 5 includes both software claims and closed-loop acquisition control.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":5,\\\"objective\\\":\\\"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\\\",\\\"rationale\\\":\\\"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":6,\\\"objective\\\":\\\"Create a structured artifact summarizing the evidence and limitations.\\\",\\\"rationale\\\":\\\"A compact evidence matrix will make the systematic survey auditable and useful.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":7,\\\"objective\\\":\\\"Submit the final technical answer with evidence-matched citations.\\\",\\\"rationale\\\":\\\"Required final delivery step.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null}]}\"}", + "name": "create_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Plan registered and formatted correctly, with some modifications:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "create_plan", + "tool_call_id": "call_TMwBuCDwZNGWGVmRuRrINX4G", + "info": { + "start_ts": 386.692433131, + "end_ts": 386.695581364 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_9At2TUTAjnJ23z8NGaLTDASo", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated optimization indexing parameters Hough Bayesian optimization genetic algorithm grid search\"}", + "name": "paper_search" + } + }, + { + "id": "call_5aBBEXJupU0KmlHg00b79lgu", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD reindex raw patterns confidence index image quality objective optimization parameter tuning\"}", + "name": "paper_search" + } + }, + { + "id": "call_acsPy2saVbrPRcVvQrHXhwqx", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD hybrid Hough dictionary indexing low confidence patterns automated\"}", + "name": "paper_search" + } + }, + { + "id": "call_SvSRR7s0yK3PpnhQQ0Vw5Xl4", + "type": "function", + "function": { + "arguments": "{\"query\": \"autonomous self-driving EBSD acquisition post-processing SEM workflow\"}", + "name": "paper_search" + } + }, + { + "id": "call_yBdeyyzRACDmvHLG4V5YZoxC", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD grain reconstruction parameter optimization misorientation threshold wild grains cleanup automated\"}", + "name": "paper_search" + } + }, + { + "id": "call_4uZ0w5ewQsn4Ra9JRMzH48rh", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated parameter optimization indexing Hough dictionary feedback\"}", + "name": "patent_search" + } + }, + { + "id": "call_WJzVeGxOMudv7uXJmeYdSe3F", + "type": "function", + "function": { + "arguments": "{\"query\": \"electron backscatter diffraction autonomous acquisition processing\"}", + "name": "patent_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Emerging materials intelligence ecosystems propelled by machine learning\nValid Text Names:\n'batra2021emergingmaterialsintelligence pages 1-2', 'batra2021emergingmaterialsintelligence pages 2-3', 'batra2021emergingmaterialsintelligence pages 3-4', 'batra2021emergingmaterialsintelligence pages 4-5', 'batra2021emergingmaterialsintelligence pages 5-6', 'batra2021emergingmaterialsintelligence pages 6-7', 'batra2021emergingmaterialsintelligence pages 7-8', 'batra2021emergingmaterialsintelligence pages 8-9', 'batra2021emergingmaterialsintelligence pages 9-10', 'batra2021emergingmaterialsintelligence pages 10-11', 'batra2021emergingmaterialsintelligence pages 11-12', 'batra2021emergingmaterialsintelligence pages 12-13', 'batra2021emergingmaterialsintelligence pages 13-14', 'batra2021emergingmaterialsintelligence pages 14-15', 'batra2021emergingmaterialsintelligence pages 15-16', 'batra2021emergingmaterialsintelligence pages 16-17', 'batra2021emergingmaterialsintelligence pages 17-18', 'batra2021emergingmaterialsintelligence pages 18-19', 'batra2021emergingmaterialsintelligence pages 19-20', 'batra2021emergingmaterialsintelligence pages 20-21', 'batra2021emergingmaterialsintelligence pages 21-22', 'batra2021emergingmaterialsintelligence pages 21-21', 'batra2021emergingmaterialsintelligence pages 22-22', 'batra2021emergingmaterialsintelligence pages 23-23', 'batra2021emergingmaterialsintelligence pages 23-24'\n\nBibTex:\n@article{batra2021emergingmaterialsintelligence,\n author = \"Batra, Rohit and Song, Le and Ramprasad, Rampi\",\n title = \"Emerging materials intelligence ecosystems propelled by machine learning\",\n year = \"2021\",\n journal = \"Nature Reviews Materials\",\n volume = \"6\",\n pages = \"655-678\",\n month = \"Nov\",\n doi = \"10.1038/s41578-020-00255-y\",\n url = \"https://doi.org/10.1038/s41578-020-00255-y\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"8\",\n issn = \"2058-8437\"\n}\n\n\nRelevant Snippet:\n\u2026 For the first point, Bayesian optimization and genetic algorithms have been successfully \u2026 than traditional indexing methods that are either susceptible to noise (Hough transformation \u2026\n\n\n---\n\n## 2. Materials Informatics for Mechanical Deformation: A Review of Applications and Challenges\nValid Text Names:\n'frydrych2021materialsinformaticsfor pages 1-2', 'frydrych2021materialsinformaticsfor pages 2-4', 'frydrych2021materialsinformaticsfor pages 4-5', 'frydrych2021materialsinformaticsfor pages 5-7', 'frydrych2021materialsinformaticsfor pages 7-8', 'frydrych2021materialsinformaticsfor pages 8-10', 'frydrych2021materialsinformaticsfor pages 10-12', 'frydrych2021materialsinformaticsfor pages 12-13', 'frydrych2021materialsinformaticsfor pages 13-15', 'frydrych2021materialsinformaticsfor pages 15-17', 'frydrych2021materialsinformaticsfor pages 17-19', 'frydrych2021materialsinformaticsfor pages 19-20', 'frydrych2021materialsinformaticsfor pages 20-22', 'frydrych2021materialsinformaticsfor pages 22-23', 'frydrych2021materialsinformaticsfor pages 23-24', 'frydrych2021materialsinformaticsfor pages 24-26', 'frydrych2021materialsinformaticsfor pages 26-27', 'frydrych2021materialsinformaticsfor pages 27-28', 'frydrych2021materialsinformaticsfor pages 28-30', 'frydrych2021materialsinformaticsfor pages 30-31'\n\nBibTex:\n@article{frydrych2021materialsinformaticsfor,\n author = \"Frydrych, Karol and Karimi, Kamran and Pecelerowicz, Michal and Alvarez, Rene and Dominguez-Guti\u00e9rrez, Francesco Javier and Rovaris, Fabrizio and Papanikolaou, Stefanos\",\n title = \"Materials Informatics for Mechanical Deformation: A Review of Applications and Challenges\",\n year = \"2021\",\n journal = \"Materials\",\n volume = \"14\",\n pages = \"5764\",\n month = \"Oct\",\n doi = \"10.3390/ma14195764\",\n url = \"https://doi.org/10.3390/ma14195764\",\n publisher = \"MDPI AG\",\n issue = \"19\",\n issn = \"1996-1944\"\n}\n\n\nAbstract:\nIn the design and development of novel materials that have excellent mechanical properties, classification and regression methods have been diversely used across mechanical deformation simulations or experiments. The use of materials informatics methods on large data that originate in experiments or/and multiscale modeling simulations may accelerate materials\u2019 discovery or develop new understanding of materials\u2019 behavior. In this fast-growing field, we focus on reviewing advances at the intersection of data science with mechanical deformation simulations and experiments, with a particular focus on studies of metals and alloys. We discuss examples of applications, as well as identify challenges and prospects.\n\nRelevant Snippet:\n\u2026 the speed of indexing and classification of EBSD patterns as \u2026 EBSD data. For example, ML methods can be used [75] to \u2026 [78] based on an evolutionary algorithm (EA) may be applied, \u2026\n\n\n---\n\n## 3. Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields\nValid Text Names:\n'stinville2022multimodaldatasetof pages 1-2', 'stinville2022multimodaldatasetof pages 2-3', 'stinville2022multimodaldatasetof pages 3-5', 'stinville2022multimodaldatasetof pages 5-7', 'stinville2022multimodaldatasetof pages 7-9', 'stinville2022multimodaldatasetof pages 9-12', 'stinville2022multimodaldatasetof pages 12-13', 'stinville2022multimodaldatasetof pages 13-15', 'stinville2022multimodaldatasetof pages 15-15', 'stinville2022multimodaldatasetof pages 15-16'\n\nBibTex:\n@article{stinville2022multimodaldatasetof,\n author = \"Stinville, J. C. and Hestroffer, J. M. and Charpagne, M. A. and Polonsky, A. T. and Echlin, M. P. and Torbet, C. J. and Valle, V. and Nygren, K. E. and Miller, M. P. and Klaas, O. and Loghin, A. and Beyerlein, I. J. and Pollock, T. M.\",\n title = \"Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields\",\n year = \"2022\",\n journal = \"Scientific Data\",\n volume = \"9\",\n month = \"Aug\",\n doi = \"10.1038/s41597-022-01525-w\",\n url = \"https://doi.org/10.1038/s41597-022-01525-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2052-4463\"\n}\n\n\nAbstract:\nAbstractThe development of high-fidelity mechanical property prediction models for the design of polycrystalline materials relies on large volumes of microstructural feature data. Concurrently, at these same scales, the deformation fields that develop during mechanical loading can be highly heterogeneous. Spatially correlated measurements of 3D microstructure and the ensuing deformation fields at the micro-scale would provide highly valuable insight into the relationship between microstructure and macroscopic mechanical response. They would also provide direct validation for numerical simulations that can guide and speed up the design of new materials and microstructures. However, to date, such data have been rare. Here, a one-of-a-kind, multi-modal dataset is presented that combines recent state-of-the-art experimental developments in 3D tomography and high-resolution deformation field measurements.\n\nRelevant Snippet:\n\u2026 maps and indexed as individual features, using the iterative Hough \u2026 Reindexing of the EBSD data with the dictionary indexing \u2026 approach using an evolutionary algorithm. Mater. Charact. \u2026\n\n\n---\n\n## 4. Bayesian Optimization for Diffraction Profile Modeling\nValid Text Names:\n'panchi2023bayesianoptimizationfor pages 1-8', 'panchi2023bayesianoptimizationfor pages 8-13', 'panchi2023bayesianoptimizationfor pages 13-18', 'panchi2023bayesianoptimizationfor pages 18-22', 'panchi2023bayesianoptimizationfor pages 22-26', 'panchi2023bayesianoptimizationfor pages 26-30', 'panchi2023bayesianoptimizationfor pages 30-34', 'panchi2023bayesianoptimizationfor pages 34-39', 'panchi2023bayesianoptimizationfor pages 39-44', 'panchi2023bayesianoptimizationfor pages 44-50', 'panchi2023bayesianoptimizationfor pages 50-54', 'panchi2023bayesianoptimizationfor pages 54-60', 'panchi2023bayesianoptimizationfor pages 60-66', 'panchi2023bayesianoptimizationfor pages 66-70', 'panchi2023bayesianoptimizationfor pages 70-74'\n\nBibTex:\n@article{panchi2023bayesianoptimizationfor,\n author = \"Panchi, Navid\",\n title = \"Bayesian Optimization for Diffraction Profile Modeling\",\n year = \"2023\",\n journal = \"Proceedings of the Institute for Multiscale Simulation\",\n month = \"Jan\",\n doi = \"10.25593/open-fau-1463\",\n url = \"https://doi.org/10.25593/open-fau-1463\",\n publisher = \"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)\"\n}\n\n\n\nUnobtainable Papers:\nDavid J Rowenhorst and William Lenthe. Ebsd pattern center fitting with particle swarm optimization. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.304, doi:10.1093/mam/ozaf048.304.\nYiling Huang, Fan Peng, Xuemei Song, Xingyu Jin, Yuqing Jiang, Wei Zheng, Caifen Jiang, Zhaoqi Wu, and Yi Zeng. Optimisation of ebsd indexing through pattern centre calibration and grain boundary refinement. Journal of microscopy, May 2025. URL: https://doi.org/10.1111/jmi.13434, doi:10.1111/jmi.13434.\nF. Briffod, T. E. J. Edwards, J. Quinta da Fonseca, J. -C. Stinville, D. Texier, and T. Vermeij. Understanding strain localization in metallic materials: a review of high-resolution digital image correlation and related techniques. Science and Technology of Advanced Materials, Mar 2026. URL: https://doi.org/10.1080/14686996.2026.2630488, doi:10.1080/14686996.2026.2630488.\nYunxuan Zhou, Zihao Wang, Wenhui Tao, Yongkang Sun, Junjie Wu, Gang Wang, Yu Xiu, Huiyu Ji, Yulin Liu, Anping Dong, Jie Wang, Jun Wang, Mengmeng Wang, and Qi Liu. Multi-objective optimization design of al-si alloys based on machine learning. Sep 2025. URL: https://doi.org/10.1016/j.calphad.2025.102870, doi:10.1016/j.calphad.2025.102870.\nZihao Ding. Applications of machine learning in crystallographic orientation determination. Text, Jan 2022. URL: https://doi.org/10.1184/r1/20750590, doi:10.1184/r1/20750590.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_9At2TUTAjnJ23z8NGaLTDASo", + "info": { + "start_ts": 397.733964229, + "end_ts": 437.781901546 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring.\nValid Text Names:\n'amini2024sustainabilityinspiredfabrication pages 1-2', 'amini2024sustainabilityinspiredfabrication pages 2-3', 'amini2024sustainabilityinspiredfabrication pages 3-5', 'amini2024sustainabilityinspiredfabrication pages 5-6', 'amini2024sustainabilityinspiredfabrication pages 6-8', 'amini2024sustainabilityinspiredfabrication pages 8-10', 'amini2024sustainabilityinspiredfabrication pages 10-11', 'amini2024sustainabilityinspiredfabrication pages 11-12', 'amini2024sustainabilityinspiredfabrication pages 12-13', 'amini2024sustainabilityinspiredfabrication pages 13-14', 'amini2024sustainabilityinspiredfabrication pages 14-15', 'amini2024sustainabilityinspiredfabrication pages 15-16', 'amini2024sustainabilityinspiredfabrication pages 16-16'\n\nBibTex:\n@article{amini2024sustainabilityinspiredfabrication,\n author = \"Amini, Shahram and Choi, Hongbin and Seche, Wesley and Blagojevic, Alexander and May, Nicholas and Lefler, Benjamin M. and Davis, Skyler L. and Elyahoodayan, Sahar and Tavousi, Pouya and May, Steven J. and Caputo, Gregory A. and Lowe, Terry C. and Hettinger, Jeffrey and Shahbazmohamadi, Sina\",\n title = \"Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring.\",\n year = \"2024\",\n journal = \"Microsystems \\& nanoengineering\",\n volume = \"10 1\",\n pages = \"125\",\n month = \"Sep\",\n doi = \"10.1038/s41378-024-00754-w\",\n url = \"https://doi.org/10.1038/s41378-024-00754-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2055-7434\"\n}\n\n\nAbstract:\nOver the last two decades, platinum group metals (PGMs) and their alloys have dominated as the materials of choice for electrodes in long-term implantable neurostimulation and cardiac rhythm management devices due to their superior conductivity, mechanical and chemical stability, biocompatibility, corrosion resistance, radiopacity, and electrochemical performance. Despite these benefits, PGM manufacturing processes are extremely costly, complex, and challenging with potential health hazards. Additionally, the volatility in PGM prices and their high supply risk, combined with their scarce concentration of approximately 0.01 ppm in the earth's upper crust and limited mining geographical areas, underscores their classification as critical raw materials, thus, their effective recovery or substitution worldwide is of paramount importance. Since postmortem recovery from deceased patients and/or refining of PGMs that are used in the manufacturing of the electrodes and microelectrode arrays is extremely rare, challenging, and highly costly, therefore, substitution of PGM-based electrodes with other biocompatible materials that can yield electrochemical performance values equal or greater than PGMs is the only viable and sustainable solution to reduce and ultimately substitute the use of PGMs in long-term implantable neurostimulation and cardiac rhythm management devices. In this article, we demonstrate for the first time how the novel technique of \"reactive hierarchical surface restructuring\" can be utilized on titanium-that is widely used in many non-stimulation medical device and implant applications-to manufacture biocompatible, low-cost, sustainable, and high-performing neurostimulation and cardiac rhythm management electrodes. We have shown how the surface of titanium electrodes with extremely poor electrochemical performance undergoes compositional and topographical transformations that result in electrodes with outstanding electrochemical performance.\n\nRelevant Snippet:\n\u2026 To achieve the surface quality required for EBSD analysis, ion milling was \u2026 pattern average re-indexing was performed on the dataset. Grain dilation and neighbor confidence index \u2026\n\n\nUnobtainable Papers:\nAli Riza Durmaz, James D. Lamb, Kamilla Zaripova, Madeline Vailhe, Julia T. Pu\u0308rstl, Justine Schulte, Marc Ackermann, McLean P. Echlin, and Tresa M. Pollock. A correlative microscopy dataset for multimodal data fusion and image matching in materials science. ChemRxiv, Apr 2026. URL: https://doi.org/10.26434/chemrxiv.15001724/v1, doi:10.26434/chemrxiv.15001724/v1.\nPatrick T. Brewick, Stuart I. Wright, and David J. Rowenhorst. Nlpar: non-local smoothing for enhanced ebsd pattern indexing. Ultramicroscopy, 200:50-61, May 2019. URL: https://doi.org/10.1016/j.ultramic.2019.02.013, doi:10.1016/j.ultramic.2019.02.013.\nCaleb E. Andrews, Maria Strantza, Nicholas P. Calta, Manyalibo J. Matthews, and Mitra L. Taheri. A denoising autoencoder for improved kikuchi pattern quality and indexing in electron backscatter diffraction. Ultramicroscopy, 253:113810, Nov 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113810, doi:10.1016/j.ultramic.2023.113810.\nA.E. Davis, X. Zeng, R. Thomas, J.R. Kennedy, J. Donoghue, A. Gholinia, P.B. Prangnell, and J. Quinta Da Fonseca. Optimising large-area crystal orientation mapping of nanoscale \u03b2 phase in \u03b1 + \u03b2 titanium alloys using ebsd. Materials Characterization, 194:112371, Dec 2022. URL: https://doi.org/10.1016/j.matchar.2022.112371, doi:10.1016/j.matchar.2022.112371.\nCatherine Zhou. Combinatorial substrate epitaxy and computationally guided epitaxial synthesis investigations of the aemno3 (ae = ca, sr, and ba) polymorphs. Text, Jan 2023. URL: https://doi.org/10.1184/r1/22313545.v1, doi:10.1184/r1/22313545.v1.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_5aBBEXJupU0KmlHg00b79lgu", + "info": { + "start_ts": 397.767522477, + "end_ts": 440.755996379 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nQiwei Shi, Dominique Loisnard, Yang Li, Zhe Chen, Haowei Wang, and St\u00e9phane Roux. Accuracy assessment of crystal orientation indexations by ebsd. Measurement Science and Technology, 35:045030, Jan 2024. URL: https://doi.org/10.1088/1361-6501/ad204d, doi:10.1088/1361-6501/ad204d.\nPatrick Trimby, Mohammed Al-Mosawi, Maisoon Al-Jawad, Stuart Micklethwaite, Zabeada Aslam, Aimo Winkelmann, and Sandra Piazolo. The characterisation of dental enamel using transmission kikuchi diffraction in the scanning electron microscope combined with dynamic template matching. Ultramicroscopy, 260:113940, Jun 2024. URL: https://doi.org/10.1016/j.ultramic.2024.113940, doi:10.1016/j.ultramic.2024.113940.\nFarangis Ram, Stuart Wright, Saransh Singh, and Marc De Graef. Error analysis of the crystal orientations obtained by the dictionary approach to ebsd indexing. Ultramicroscopy, 181:17-26, Oct 2017. URL: https://doi.org/10.1016/j.ultramic.2017.04.016, doi:10.1016/j.ultramic.2017.04.016.\nKatharina Marquardt, Marc De Graef, Saransh Singh, Hauke Marquardt, Anja Rosenthal, and Sanae Koizuimi. Quantitative electron backscatter diffraction (ebsd) data analyses using the dictionary indexing (di) approach: overcoming indexing difficulties on geological materials. American Mineralogist, 102:1843-1855, Sep 2017. URL: https://doi.org/10.2138/am-2017-6062, doi:10.2138/am-2017-6062.\nDavid J. Rowenhorst, Patrick G. Callahan, and H\u00e5kon W. \u00c5nes. Fast radon transforms for high-precision ebsd orientation determination using pyebsdindex. Journal of Applied Crystallography, 57:3-19, Feb 2024. URL: https://doi.org/10.1107/s1600576723010221, doi:10.1107/s1600576723010221.\nK. V. Mani Krishna, R. Madhavan, Mangesh V. Pantawane, Rajarshi Banerjee, and Narendra B. Dahotre. Machine learning based de-noising of electron back scatter patterns of various crystallographic metallic materials fabricated using laser directed energy deposition. Ultramicroscopy, 247:113703, May 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113703, doi:10.1016/j.ultramic.2023.113703.\nM. A. Jackson, E. Pascal, and M. De Graef. Dictionary indexing of electron back-scatter diffraction patterns: a hands-on tutorial. Integrating Materials and Manufacturing Innovation, 8:226-246, May 2019. URL: https://doi.org/10.1007/s40192-019-00137-4, doi:10.1007/s40192-019-00137-4.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_acsPy2saVbrPRcVvQrHXhwqx", + "info": { + "start_ts": 397.799029707, + "end_ts": 471.298576361 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Recent advances and applications of deep learning methods in materials science\nValid Text Names:\n'choudhary2110recentadvancesand pages 1-2', 'choudhary2110recentadvancesand pages 2-3', 'choudhary2110recentadvancesand pages 3-4', 'choudhary2110recentadvancesand pages 3-3', 'choudhary2110recentadvancesand pages 4-5', 'choudhary2110recentadvancesand pages 5-6', 'choudhary2110recentadvancesand pages 6-8', 'choudhary2110recentadvancesand pages 8-10', 'choudhary2110recentadvancesand pages 8-8', 'choudhary2110recentadvancesand pages 10-11', 'choudhary2110recentadvancesand pages 11-12', 'choudhary2110recentadvancesand pages 11-11', 'choudhary2110recentadvancesand pages 14-15', 'choudhary2110recentadvancesand pages 16-17', 'choudhary2110recentadvancesand pages 17-18', 'choudhary2110recentadvancesand pages 18-19', 'choudhary2110recentadvancesand pages 19-20', 'choudhary2110recentadvancesand pages 20-21', 'choudhary2110recentadvancesand pages 20-20', 'choudhary2110recentadvancesand pages 21-22', 'choudhary2110recentadvancesand pages 22-23', 'choudhary2110recentadvancesand pages 22-22', 'choudhary2110recentadvancesand pages 23-24', 'choudhary2110recentadvancesand pages 24-24', 'choudhary2110recentadvancesand pages 25-26'\n\nBibTex:\n@article{choudhary2022recentadvancesand,\n author = \"Choudhary, Kamal and DeCost, Brian and Chen, Chi and Jain, Anubhav and Tavazza, Francesca and Cohn, Ryan and Park, Cheol Woo and Choudhary, Alok and Agrawal, Ankit and Billinge, Simon J. L. and Holm, Elizabeth and Ong, Shyue Ping and Wolverton, Chris\",\n title = \"Recent advances and applications of deep learning methods in materials science\",\n year = \"2022\",\n journal = \"npj Computational Materials\",\n volume = \"8\",\n month = \"Apr\",\n doi = \"10.1038/s41524-022-00734-6\",\n url = \"https://doi.org/10.1038/s41524-022-00734-6\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\",\n pages = \"1-26\"\n}\n\n\nAbstract:\nAbstractDeep learning (DL) is one of the fastest-growing topics in materials data science, with rapidly emerging applications spanning atomistic, image-based, spectral, and textual data modalities. DL allows analysis of unstructured data and automated identification of features. The recent development of large materials databases has fueled the application of DL methods in atomistic prediction in particular. In contrast, advances in image and spectral data have largely leveraged synthetic data enabled by high-quality forward models as well as by generative unsupervised DL methods. In this article, we present a high-level overview of deep learning methods followed by a detailed discussion of recent developments of deep learning in atomistic simulation, materials imaging, spectral analysis, and natural language processing. For each modality we discuss applications involving both theoretical and experimental data, typical modeling approaches with their strengths and limitations, and relevant publicly available software and datasets. We conclude the review with a discussion of recent cross-cutting work related to uncertainty quantification in this field and a brief perspective on limitations, challenges, and potential growth areas for DL methods in materials science.\n\nRelevant Snippet:\n\u2026 EBSD indexing capable of taking an EBSD image as input and predicting the three Euler angles representing the orientation that would have led to the given EBSD \u2026 post-processing of \u2026\n\n\nUnobtainable Papers:\nGrace Guinan, Addison Salvador, Michelle A. Smeaton, Andrew Glaws, Hilary Egan, Brian C. Wyatt, Babak Anasori, Kevin R. Fiedler, Matthew J. Olszta, and Steven R. Spurgeon. Mind the gap: bridging the divide between ai aspirations and the reality of autonomous microscopy. Jun 2025. URL: https://doi.org/10.1063/5.0267699, doi:10.1063/5.0267699.\nMohammad Mazedul Islam, T. M. A. Iqbal Bin Belal, Md. Arif Hossain Chowdhury Anik, and Ahmed Sharif. Predicting materials using variational autoencoders: a systematic literature review. Archives of Computational Methods in Engineering, 33(5):6599-6650, Jan 2026. URL: https://doi.org/10.1007/s11831-025-10485-4, doi:10.1007/s11831-025-10485-4.\nChunyu Guo, Xiangyu Tang, Yu\u2019e Chen, Changyou Gao, Qinglin Shan, Heyi Wei, Xusheng Liu, Chuncheng Lu, Meixia Fu, Enhui Wang, Xinhong Liu, Xinmei Hou, and Yanglong Hou. From microstructure to performance optimization: innovative applications of computer vision in materials science. International Journal of Minerals, Metallurgy and Materials, 33(1):94-115, Dec 2026. URL: https://doi.org/10.1007/s12613-025-3217-4, doi:10.1007/s12613-025-3217-4.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_SvSRR7s0yK3PpnhQQ0Vw5Xl4", + "info": { + "start_ts": 397.833985678, + "end_ts": 435.362992784 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\nValid Text Names:\n'polonsky2019threedimensionalanalysisand pages 1-2', 'polonsky2019threedimensionalanalysisand pages 2-5', 'polonsky2019threedimensionalanalysisand pages 5-6', 'polonsky2019threedimensionalanalysisand pages 6-8', 'polonsky2019threedimensionalanalysisand pages 8-11', 'polonsky2019threedimensionalanalysisand pages 11-13', 'polonsky2019threedimensionalanalysisand pages 13-14', 'polonsky2019threedimensionalanalysisand pages 14-15', 'polonsky2019threedimensionalanalysisand pages 15-15'\n\nBibTex:\n@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a cloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline variables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore, a comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using BisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects for impact in material science are also discussed.\n\nRelevant Snippet:\n\u2026 about the grain segmentation misorientation tolerance in \u2026 Automated orientation imaging analysis via EBSD first \u2026 During the reconstruction process, various cleanup steps are employed \u2026\n\n\n---\n\n## 2. 3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry\nValid Text Names:\n'sparks20243dreconstructionof pages 1-2', 'sparks20243dreconstructionof pages 2-6', 'sparks20243dreconstructionof pages 6-7', 'sparks20243dreconstructionof pages 7-9', 'sparks20243dreconstructionof pages 9-12', 'sparks20243dreconstructionof pages 12-14', 'sparks20243dreconstructionof pages 14-16', 'sparks20243dreconstructionof pages 16-18', 'sparks20243dreconstructionof pages 18-19', 'sparks20243dreconstructionof pages 19-21', 'sparks20243dreconstructionof pages 21-23', 'sparks20243dreconstructionof pages 23-24', 'sparks20243dreconstructionof pages 24-26', 'sparks20243dreconstructionof pages 26-28', 'sparks20243dreconstructionof pages 28-30', 'sparks20243dreconstructionof pages 30-31', 'sparks20243dreconstructionof pages 31-31'\n\nBibTex:\n@article{sparks20243dreconstructionof,\n author = \"Sparks, Gregory and Mason, Simon A. and Chapman, Michael G. and Park, Jun-Sang and Sharma, Hemant and Kenesei, Peter and Niezgoda, Stephen R. and Mills, Michael J. and Uchic, Michael D. and Shade, Paul A. and Obstalecki, Mark\",\n title = \"3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jul\",\n doi = \"10.1007/s40192-024-00370-6\",\n url = \"https://doi.org/10.1007/s40192-024-00370-6\",\n volume = \"13\",\n pages = \"773-803\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"3\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 EBSD, pattern center fitting for every section, automated \u2026 Grains were segmented using a misorientation threshold of 3 \u25e6\u2026 3D pipelines that were used for cleanup and segmentation, \u2026\n\n\n---\n\n## 3. Recent Progress of Digital Reconstruction in Polycrystalline Materials\nValid Text Names:\n'chen2025recentprogressof pages 1-2', 'chen2025recentprogressof pages 2-4', 'chen2025recentprogressof pages 4-7', 'chen2025recentprogressof pages 10-12', 'chen2025recentprogressof pages 12-15', 'chen2025recentprogressof pages 15-18', 'chen2025recentprogressof pages 18-19', 'chen2025recentprogressof pages 19-20', 'chen2025recentprogressof pages 20-24', 'chen2025recentprogressof pages 24-25', 'chen2025recentprogressof pages 25-27', 'chen2025recentprogressof pages 27-28', 'chen2025recentprogressof pages 28-32', 'chen2025recentprogressof pages 32-34', 'chen2025recentprogressof pages 34-36', 'chen2025recentprogressof pages 38-40', 'chen2025recentprogressof pages 41-42', 'chen2025recentprogressof pages 42-43', 'chen2025recentprogressof pages 43-44', 'chen2025recentprogressof pages 45-46', 'chen2025recentprogressof pages 46-47', 'chen2025recentprogressof pages 47-48', 'chen2025recentprogressof pages 48-49', 'chen2025recentprogressof pages 49-50', 'chen2025recentprogressof pages 50-51'\n\nBibTex:\n@article{chen2025recentprogressof,\n author = \"Chen, Bingbing and Li, Dongfeng and Davies, Peter and Johnston, Richard and Ge, Xiangyun and Li, Chenfeng\",\n title = \"Recent Progress of Digital Reconstruction in Polycrystalline Materials\",\n year = \"2025\",\n journal = \"Archives of Computational Methods in Engineering\",\n month = \"Mar\",\n doi = \"10.1007/s11831-025-10245-4\",\n url = \"https://doi.org/10.1007/s11831-025-10245-4\"\n}\n\n\nAbstract:\n\n This study comprehensively reviews recent advances in the digital reconstruction of polycrystalline materials. Digital reconstruction serves as both a representative volume element for multiscale modelling and a source of quantitative data for microstructure characterisation. Three main types of digital reconstruction in polycrystalline materials exist: (i) experimental reconstruction, which links processing-structure-properties-performance by reconstructing actual polycrystalline microstructures using destructive or non-destructive methods; (ii) physics-based models, which replicate evolutionary processes to establish processing-structure linkages, including cellular automata, Monte Carlo, vertex/front tracking, level set, machine learning, and phase field methods; and (iii) geometry-based models, which create ensembles of statistically equivalent polycrystalline microstructures for structure-properties-performance linkages, using simplistic morphology, Voronoi tessellation, ellipsoid packing, texture synthesis, high-order, reduced-order, and machine learning methods. This work reviews the key features, procedures, advantages, and limitations of these methods, with a particular focus on their application in constructing processing-structure-properties-performance linkages. Finally, it summarises the conclusions, challenges, and future directions for digital reconstruction in polycrystalline materials within the framework of computational materials engineering.\n\nRelevant Snippet:\n\u2026 Several automated devices are commonly used for serial \u2026 data mapping for identifying grains and twins in EBSD (Fig 3c). For \u2026 faster than GFIB [33] and subsequent surface cleanup at \u2026\n\n\nUnobtainable Papers:\nJiwoong Kang. Kinetics and connectivity of grain boundaries in three-dimensional metallic systems. Text, Jan 2022. URL: https://doi.org/10.7302/6025, doi:10.7302/6025.\nMark Taylor, Albert D Smith, Jack M Donoghue, Timothy L Burnett, and Ed Pickering. In-situ heating-stage ebsd validation of algorithms for prior-austenite grain reconstruction in steel. Scripta Materialia, Mar 2024. URL: https://doi.org/10.1016/j.scriptamat.2023.115924, doi:10.1016/j.scriptamat.2023.115924.\nAlistair Garner, Jack Donoghue, Remco Geurts, Yasser Al Aboura, Bart\u0142omiej Winiarski, Phil B. Prangnell, and Tim L. Burnett. Large-scale serial sectioning of environmentally assisted cracks in 7xxx al alloys using femtosecond laser-pfib. Materials Characterization, 188:111890, Jun 2022. URL: https://doi.org/10.1016/j.matchar.2022.111890, doi:10.1016/j.matchar.2022.111890.\nH\u00e5kon W. \u00c5nes, Antonius T.J. van Helvoort, and Knut Marthinsen. Orientation dependent pinning of (sub)grains by dispersoids during recovery and recrystallization in an al\u2013mn alloy. Acta Materialia, 248:118761, Apr 2023. URL: https://doi.org/10.1016/j.actamat.2023.118761, doi:10.1016/j.actamat.2023.118761.\nH. Latuner, Julie Favre, A. Helstroffer, G. Inacio Da Rosa, P. Joly, C. Maurice, E. Plancher, M. Roth, C. Desrayaud, and G. Kermouche. Identification of recrystallized grains in ebsd maps: automated adaptive thresholding using a gaussian mixture approach. Materials Characterization, 231:115920, Jan 2026. URL: https://doi.org/10.1016/j.matchar.2025.115920, doi:10.1016/j.matchar.2025.115920.\nThomas Girerd, Giovanna Mart\u00ednez-Arellano, Adam Thomas Clare, and Alistair Speidel. A facile methodology to identify microstructural grains on etched surfaces using panoptic segmentation. Sep 2025. URL: https://doi.org/10.1016/j.matchar.2025.115224, doi:10.1016/j.matchar.2025.115224.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_yBdeyyzRACDmvHLG4V5YZoxC", + "info": { + "start_ts": 397.873479618, + "end_ts": 440.096785711 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional patents. This may be because nothing matched the search query, or because all found patents have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=0 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "patent_search", + "tool_call_id": "call_4uZ0w5ewQsn4Ra9JRMzH48rh", + "info": { + "start_ts": 397.913352854, + "end_ts": 399.574481165 + } + }, + { + "role": "tool", + "content": "Retrieved Patents:\n## 1. Optical package for a lidar sensor system and lidar sensor system technical \u2026\nValid Text Names:\n'US20220276352A1 pages 10-18', 'US20220276352A1 pages 24-25', 'US20220276352A1 pages 25-26', 'US20220276352A1 pages 26-27', 'US20220276352A1 pages 27-28', 'US20220276352A1 pages 29-30', 'US20220276352A1 pages 30-31', 'US20220276352A1 pages 31-32', 'US20220276352A1 pages 32-33', 'US20220276352A1 pages 34-35', 'US20220276352A1 pages 37-38', 'US20220276352A1 pages 38-39', 'US20220276352A1 pages 39-40', 'US20220276352A1 pages 41-42', 'US20220276352A1 pages 42-43', 'US20220276352A1 pages 44-45', 'US20220276352A1 pages 46-46', 'US20220276352A1 pages 47-48', 'US20220276352A1 pages 48-49', 'US20220276352A1 pages 49-50', 'US20220276352A1 pages 52-53', 'US20220276352A1 pages 58-59', 'US20220276352A1 pages 59-60', 'US20220276352A1 pages 60-61', 'US20220276352A1 pages 61-62'\n\nBibTex:\n@article{US20220276352A1,\n author = \"Ferreira, Ricardo\",\n title = \"Optical package for a lidar sensor system and lidar sensor system technical \u2026\",\n year = \"2022\",\n journal = \"Patent (WO,EP,US,CN,CA,DE)\"\n}\n\n\nRelevant Snippet:\nThe present disclosure relates to various embodiments of an optical package for a LIDAR sensor system having a substrate including an array of a plurality of capacitors formed in the substrate, a plurality of switches formed in the substrate, where each switch is connected between at least one \u2026\n\n\n---\n\n## 2. Hyperspectral sensing system and methods\nValid Text Names:\n'US11650190B2 pages 1-4', 'US11650190B2 pages 4-6', 'US11650190B2 pages 6-9', 'US11650190B2 pages 12-15', 'US11650190B2 pages 18-21', 'US11650190B2 pages 26-29', 'US11650190B2 pages 29-31', 'US11650190B2 pages 31-34', 'US11650190B2 pages 34-37', 'US11650190B2 pages 39-41', 'US11650190B2 pages 41-44', 'US11650190B2 pages 46-49', 'US11650190B2 pages 53-56', 'US11650190B2 pages 58-59', 'US11650190B2 pages 61-63', 'US11650190B2 pages 65-67', 'US11650190B2 pages 69-71', 'US11650190B2 pages 81-83', 'US11650190B2 pages 87-89', 'US11650190B2 pages 89-91', 'US11650190B2 pages 96-98', 'US11650190B2 pages 100-102', 'US11650190B2 pages 104-106', 'US11650190B2 pages 106-108', 'US11650190B2 pages 110-112'\n\nBibTex:\n@article{US11650190B2,\n author = \"Tufillaro, Nicholas\",\n title = \"Hyperspectral sensing system and methods\",\n year = \"2023\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nA method for retrieving a corrected spectrum from a measured spectrum (e.g., retrieving a top-of-water spectrum from a measured top-of-atmosphere spectrum) includes creating a scene-specific model of a region of interest and performing a ray-tracing simulation to simulate rays of light that would \u2026\n\n\n---\n\n## 3. Multidimensional materials sensing systems and methods\nValid Text Names:\n'US12517250B2 pages 1-4', 'US12517250B2 pages 4-6', 'US12517250B2 pages 6-9', 'US12517250B2 pages 9-11', 'US12517250B2 pages 11-14', 'US12517250B2 pages 14-17', 'US12517250B2 pages 19-22', 'US12517250B2 pages 22-25', 'US12517250B2 pages 33-35', 'US12517250B2 pages 38-41', 'US12517250B2 pages 41-43', 'US12517250B2 pages 43-46', 'US12517250B2 pages 46-48', 'US12517250B2 pages 48-51', 'US12517250B2 pages 51-53', 'US12517250B2 pages 53-55', 'US12517250B2 pages 55-57', 'US12517250B2 pages 59-61', 'US12517250B2 pages 61-62', 'US12517250B2 pages 64-66', 'US12517250B2 pages 66-68', 'US12517250B2 pages 68-70', 'US12517250B2 pages 74-76', 'US12517250B2 pages 78-80', 'US12517250B2 pages 86-86'\n\nBibTex:\n@article{US12517250B2,\n author = \"Kotov, Nicholas A.\",\n title = \"Multidimensional materials sensing systems and methods\",\n year = \"2026\",\n journal = \"Patent (WO,EP,US,CN,JP,KR)\"\n}\n\n\nRelevant Snippet:\nMultidimensional Light Imaging, Detection, and Ranging (LIDAR) systems and methods optionally include a laser device configured to generate a plurality of light pulses emitted towards an object, a detector configured to receive a portion of the plurality of light pulses returned from the object, \u2026\n\n\n---\n\n## 4. Super system on chip\nValid Text Names:\n'US12380059B2 pages 1-4', 'US12380059B2 pages 18-20', 'US12380059B2 pages 20-23', 'US12380059B2 pages 23-26', 'US12380059B2 pages 29-32', 'US12380059B2 pages 37-40', 'US12380059B2 pages 43-46', 'US12380059B2 pages 59-62', 'US12380059B2 pages 75-78', 'US12380059B2 pages 86-89', 'US12380059B2 pages 89-92', 'US12380059B2 pages 100-103', 'US12380059B2 pages 111-114', 'US12380059B2 pages 125-128', 'US12380059B2 pages 137-140', 'US12380059B2 pages 148-150', 'US12380059B2 pages 154-156', 'US12380059B2 pages 156-158', 'US12380059B2 pages 185-187', 'US12380059B2 pages 199-201', 'US12380059B2 pages 203-205', 'US12380059B2 pages 207-210', 'US12380059B2 pages 214-216', 'US12380059B2 pages 218-220', 'US12380059B2 pages 239-241'\n\nBibTex:\n@article{US12380059B2,\n author = \"Mazed, Mohammad A.\",\n title = \"Super system on chip\",\n year = \"2025\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nA Super System is disclosed and its inputs/outputs are coupled with a Mach-Zehnder interferometer (MZI), wherein the Mach-Zehnder interferometer (MZI) can be coupled with a first optical waveguide either in a two-dimensional (2-D) or in a three-dimensional (3-D) arrangement. The first optical \u2026\n\n\n---\n\n## 5. Super system on chip\nValid Text Names:\n'US11320588B1 pages 1-4', 'US11320588B1 pages 4-7', 'US11320588B1 pages 7-11', 'US11320588B1 pages 11-14', 'US11320588B1 pages 17-20', 'US11320588B1 pages 22-25', 'US11320588B1 pages 25-28', 'US11320588B1 pages 31-34', 'US11320588B1 pages 36-39', 'US11320588B1 pages 48-50', 'US11320588B1 pages 50-53', 'US11320588B1 pages 59-62', 'US11320588B1 pages 65-67', 'US11320588B1 pages 67-70', 'US11320588B1 pages 85-87', 'US11320588B1 pages 92-95', 'US11320588B1 pages 95-98', 'US11320588B1 pages 98-100', 'US11320588B1 pages 103-106', 'US11320588B1 pages 106-109', 'US11320588B1 pages 109-112', 'US11320588B1 pages 115-117', 'US11320588B1 pages 117-120', 'US11320588B1 pages 120-122', 'US11320588B1 pages 197-200'\n\nBibTex:\n@article{US11320588B1,\n author = \"Mazed, Mohammad A.\",\n title = \"Super system on chip\",\n year = \"2022\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nA Super System on Chip (SSoC) coupled with a photonic neural learning processor (PNLP), one or more quantum bits (qubits) and a machine learning algorithm for ultrafast data processing, image processing/recognition, deep learning/meta-learning and self-learning is disclosed. The Super System on \u2026\n\n\n---\n\n## 6. Methods and apparatus for three-dimensional (3D) imaging\nValid Text Names:\n'US11112503B2 pages 1-2', 'US11112503B2 pages 2-5', 'US11112503B2 pages 5-12', 'US11112503B2 pages 12-15', 'US11112503B2 pages 15-19', 'US11112503B2 pages 19-20', 'US11112503B2 pages 20-21', 'US11112503B2 pages 21-22', 'US11112503B2 pages 22-23', 'US11112503B2 pages 23-24', 'US11112503B2 pages 24-25', 'US11112503B2 pages 25-25', 'US11112503B2 pages 25-26', 'US11112503B2 pages 26-27', 'US11112503B2 pages 27-28', 'US11112503B2 pages 28-29', 'US11112503B2 pages 29-30', 'US11112503B2 pages 30-31', 'US11112503B2 pages 31-32', 'US11112503B2 pages 32-33', 'US11112503B2 pages 33-34', 'US11112503B2 pages 34-35', 'US11112503B2 pages 35-36', 'US11112503B2 pages 36-37', 'US11112503B2 pages 37-38'\n\nBibTex:\n@article{US11112503B2,\n author = \"Fried, Dale G.\",\n title = \"Methods and apparatus for three-dimensional (3D) imaging\",\n year = \"2021\",\n journal = \"Patent (WO,EP,US,CA)\"\n}\n\n\nRelevant Snippet:\nA method of imaging a scene includes estimating multiple three-dimensional (3D) representations, each of which corresponds to a respective portion of the scene. Neighboring portions of the scene area are at least partially overlapping. Each 3D representation is estimated by illuminating the \u2026\n\n\n---\n\n## 7. Smart device\nValid Text Names:\n'US10046229B2 pages 1-2', 'US10046229B2 pages 2-4', 'US10046229B2 pages 13-17', 'US10046229B2 pages 17-18', 'US10046229B2 pages 20-21', 'US10046229B2 pages 21-22', 'US10046229B2 pages 23-24', 'US10046229B2 pages 24-25', 'US10046229B2 pages 28-29', 'US10046229B2 pages 29-30', 'US10046229B2 pages 30-30', 'US10046229B2 pages 31-32', 'US10046229B2 pages 34-35', 'US10046229B2 pages 36-37', 'US10046229B2 pages 38-39', 'US10046229B2 pages 40-41', 'US10046229B2 pages 44-45', 'US10046229B2 pages 47-48', 'US10046229B2 pages 49-50', 'US10046229B2 pages 50-51', 'US10046229B2 pages 53-54', 'US10046229B2 pages 54-55', 'US10046229B2 pages 55-56', 'US10046229B2 pages 60-61', 'US10046229B2 pages 62-63'\n\nBibTex:\n@article{US10046229B2,\n author = \"Tran, Bao\",\n title = \"Smart device\",\n year = \"2018\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nAn Internet of Thing (IoT) device includes a body with a processor, a camera and a wireless transceiver coupled to the processor.\n\n\nUnobtainable Patents:\nJiwoong Kang. Kinetics and connectivity of grain boundaries in three-dimensional metallic systems. Text, Jan 2022. URL: https://doi.org/10.7302/6025, doi:10.7302/6025.\nQiwei Shi, Dominique Loisnard, Yang Li, Zhe Chen, Haowei Wang, and St\u00e9phane Roux. Accuracy assessment of crystal orientation indexations by ebsd. Measurement Science and Technology, 35:045030, Jan 2024. URL: https://doi.org/10.1088/1361-6501/ad204d, doi:10.1088/1361-6501/ad204d.\nPatrick Trimby, Mohammed Al-Mosawi, Maisoon Al-Jawad, Stuart Micklethwaite, Zabeada Aslam, Aimo Winkelmann, and Sandra Piazolo. The characterisation of dental enamel using transmission kikuchi diffraction in the scanning electron microscope combined with dynamic template matching. Ultramicroscopy, 260:113940, Jun 2024. URL: https://doi.org/10.1016/j.ultramic.2024.113940, doi:10.1016/j.ultramic.2024.113940.\nYiling Huang, Fan Peng, Xuemei Song, Xingyu Jin, Yuqing Jiang, Wei Zheng, Caifen Jiang, Zhaoqi Wu, and Yi Zeng. Optimisation of ebsd indexing through pattern centre calibration and grain boundary refinement. Journal of microscopy, May 2025. URL: https://doi.org/10.1111/jmi.13434, doi:10.1111/jmi.13434.\nDavid J. Rowenhorst, Patrick G. Callahan, and H\u00e5kon W. \u00c5nes. Fast radon transforms for high-precision ebsd orientation determination using pyebsdindex. Journal of Applied Crystallography, 57:3-19, Feb 2024. URL: https://doi.org/10.1107/s1600576723010221, doi:10.1107/s1600576723010221.\nGrace Guinan, Addison Salvador, Michelle A. Smeaton, Andrew Glaws, Hilary Egan, Brian C. Wyatt, Babak Anasori, Kevin R. Fiedler, Matthew J. Olszta, and Steven R. Spurgeon. Mind the gap: bridging the divide between ai aspirations and the reality of autonomous microscopy. Jun 2025. URL: https://doi.org/10.1063/5.0267699, doi:10.1063/5.0267699.\nMark Taylor, Albert D Smith, Jack M Donoghue, Timothy L Burnett, and Ed Pickering. In-situ heating-stage ebsd validation of algorithms for prior-austenite grain reconstruction in steel. Scripta Materialia, Mar 2024. URL: https://doi.org/10.1016/j.scriptamat.2023.115924, doi:10.1016/j.scriptamat.2023.115924.\nCaleb E. Andrews, Maria Strantza, Nicholas P. Calta, Manyalibo J. Matthews, and Mitra L. Taheri. A denoising autoencoder for improved kikuchi pattern quality and indexing in electron backscatter diffraction. Ultramicroscopy, 253:113810, Nov 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113810, doi:10.1016/j.ultramic.2023.113810.\nA.E. Davis, X. Zeng, R. Thomas, J.R. Kennedy, J. Donoghue, A. Gholinia, P.B. Prangnell, and J. Quinta Da Fonseca. Optimising large-area crystal orientation mapping of nanoscale \u03b2 phase in \u03b1 + \u03b2 titanium alloys using ebsd. Materials Characterization, 194:112371, Dec 2022. URL: https://doi.org/10.1016/j.matchar.2022.112371, doi:10.1016/j.matchar.2022.112371.\nThomas Girerd, Giovanna Mart\u00ednez-Arellano, Adam Thomas Clare, and Alistair Speidel. A facile methodology to identify microstructural grains on etched surfaces using panoptic segmentation. Sep 2025. URL: https://doi.org/10.1016/j.matchar.2025.115224, doi:10.1016/j.matchar.2025.115224.\nF. Briffod, T. E. J. Edwards, J. Quinta da Fonseca, J. -C. Stinville, D. Texier, and T. Vermeij. Understanding strain localization in metallic materials: a review of high-resolution digital image correlation and related techniques. Science and Technology of Advanced Materials, Mar 2026. URL: https://doi.org/10.1080/14686996.2026.2630488, doi:10.1080/14686996.2026.2630488.\nH. Latuner, Julie Favre, A. Helstroffer, G. Inacio Da Rosa, P. Joly, C. Maurice, E. Plancher, M. Roth, C. Desrayaud, and G. Kermouche. Identification of recrystallized grains in ebsd maps: automated adaptive thresholding using a gaussian mixture approach. Materials Characterization, 231:115920, Jan 2026. URL: https://doi.org/10.1016/j.matchar.2025.115920, doi:10.1016/j.matchar.2025.115920.\nCatherine Zhou. Combinatorial substrate epitaxy and computationally guided epitaxial synthesis investigations of the aemno3 (ae = ca, sr, and ba) polymorphs. Text, Jan 2023. URL: https://doi.org/10.1184/r1/22313545.v1, doi:10.1184/r1/22313545.v1.\nDavid J Rowenhorst and William Lenthe. Ebsd pattern center fitting with particle swarm optimization. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.304, doi:10.1093/mam/ozaf048.304.\nAli Riza Durmaz, James D. Lamb, Kamilla Zaripova, Madeline Vailhe, Julia T. Pu\u0308rstl, Justine Schulte, Marc Ackermann, McLean P. Echlin, and Tresa M. Pollock. A correlative microscopy dataset for multimodal data fusion and image matching in materials science. ChemRxiv, Apr 2026. URL: https://doi.org/10.26434/chemrxiv.15001724/v1, doi:10.26434/chemrxiv.15001724/v1.\nKatharina Marquardt, Marc De Graef, Saransh Singh, Hauke Marquardt, Anja Rosenthal, and Sanae Koizuimi. Quantitative electron backscatter diffraction (ebsd) data analyses using the dictionary indexing (di) approach: overcoming indexing difficulties on geological materials. American Mineralogist, 102:1843-1855, Sep 2017. URL: https://doi.org/10.2138/am-2017-6062, doi:10.2138/am-2017-6062.\nFarangis Ram, Stuart Wright, Saransh Singh, and Marc De Graef. Error analysis of the crystal orientations obtained by the dictionary approach to ebsd indexing. Ultramicroscopy, 181:17-26, Oct 2017. URL: https://doi.org/10.1016/j.ultramic.2017.04.016, doi:10.1016/j.ultramic.2017.04.016.\nPatrick T. Brewick, Stuart I. Wright, and David J. Rowenhorst. Nlpar: non-local smoothing for enhanced ebsd pattern indexing. Ultramicroscopy, 200:50-61, May 2019. URL: https://doi.org/10.1016/j.ultramic.2019.02.013, doi:10.1016/j.ultramic.2019.02.013.\nZihao Ding. Applications of machine learning in crystallographic orientation determination. Text, Jan 2022. URL: https://doi.org/10.1184/r1/20750590, doi:10.1184/r1/20750590.\nK. V. Mani Krishna, R. Madhavan, Mangesh V. Pantawane, Rajarshi Banerjee, and Narendra B. Dahotre. Machine learning based de-noising of electron back scatter patterns of various crystallographic metallic materials fabricated using laser directed energy deposition. Ultramicroscopy, 247:113703, May 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113703, doi:10.1016/j.ultramic.2023.113703.\nAlistair Garner, Jack Donoghue, Remco Geurts, Yasser Al Aboura, Bart\u0142omiej Winiarski, Phil B. Prangnell, and Tim L. Burnett. Large-scale serial sectioning of environmentally assisted cracks in 7xxx al alloys using femtosecond laser-pfib. Materials Characterization, 188:111890, Jun 2022. URL: https://doi.org/10.1016/j.matchar.2022.111890, doi:10.1016/j.matchar.2022.111890.\nMohammad Mazedul Islam, T. M. A. Iqbal Bin Belal, Md. Arif Hossain Chowdhury Anik, and Ahmed Sharif. Predicting materials using variational autoencoders: a systematic literature review. Archives of Computational Methods in Engineering, 33(5):6599-6650, Jan 2026. URL: https://doi.org/10.1007/s11831-025-10485-4, doi:10.1007/s11831-025-10485-4.\nYunxuan Zhou, Zihao Wang, Wenhui Tao, Yongkang Sun, Junjie Wu, Gang Wang, Yu Xiu, Huiyu Ji, Yulin Liu, Anping Dong, Jie Wang, Jun Wang, Mengmeng Wang, and Qi Liu. Multi-objective optimization design of al-si alloys based on machine learning. Sep 2025. URL: https://doi.org/10.1016/j.calphad.2025.102870, doi:10.1016/j.calphad.2025.102870.\nM. A. Jackson, E. Pascal, and M. De Graef. Dictionary indexing of electron back-scatter diffraction patterns: a hands-on tutorial. Integrating Materials and Manufacturing Innovation, 8:226-246, May 2019. URL: https://doi.org/10.1007/s40192-019-00137-4, doi:10.1007/s40192-019-00137-4.\nH\u00e5kon W. \u00c5nes, Antonius T.J. van Helvoort, and Knut Marthinsen. Orientation dependent pinning of (sub)grains by dispersoids during recovery and recrystallization in an al\u2013mn alloy. Acta Materialia, 248:118761, Apr 2023. URL: https://doi.org/10.1016/j.actamat.2023.118761, doi:10.1016/j.actamat.2023.118761.\nChunyu Guo, Xiangyu Tang, Yu\u2019e Chen, Changyou Gao, Qinglin Shan, Heyi Wei, Xusheng Liu, Chuncheng Lu, Meixia Fu, Enhui Wang, Xinhong Liu, Xinmei Hou, and Yanglong Hou. From microstructure to performance optimization: innovative applications of computer vision in materials science. International Journal of Minerals, Metallurgy and Materials, 33(1):94-115, Dec 2026. URL: https://doi.org/10.1007/s12613-025-3217-4, doi:10.1007/s12613-025-3217-4.\n\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "patent_search", + "tool_call_id": "call_WJzVeGxOMudv7uXJmeYdSe3F", + "info": { + "start_ts": 397.955491698, + "end_ts": 607.461565427 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_uh7IvP3yQ85eCVKwBxCbjRkg", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD \\\"evolutionary algorithm\\\" indexing pattern matching\"}", + "name": "paper_search" + } + }, + { + "id": "call_PGusrK3loi2SP0YX9X3dJxyb", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD \\\"parameter optimization\\\" Hough transform indexing\"}", + "name": "paper_search" + } + }, + { + "id": "call_KI8Kw21UgTxchIJAACgrtWy6", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD pattern center optimization differential evolution particle swarm\"}", + "name": "paper_search" + } + }, + { + "id": "call_AFeXRuxf8pxFGq9DkB0yS3lq", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD dictionary indexing low confidence Hough reindexing\"}", + "name": "paper_search" + } + }, + { + "id": "call_rjeQSC8BesdXWTotR8LKnLxa", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated acquisition dwell time beam current optimization pattern quality\"}", + "name": "paper_search" + } + }, + { + "id": "call_x3Q5do7NWtrRZiM5mvwykX6y", + "type": "function", + "function": { + "arguments": "{\"query\": \"DREAM.3D parameter sweep grain segmentation misorientation tolerance BisQue EBSD\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Effects of SEM Distortions on EBSD Grain Size Measurements\nValid Text Names:\n'timberlake2026effectsofsem pages 1-2', 'timberlake2026effectsofsem pages 2-4', 'timberlake2026effectsofsem pages 4-6', 'timberlake2026effectsofsem pages 6-8', 'timberlake2026effectsofsem pages 8-14', 'timberlake2026effectsofsem pages 14-17', 'timberlake2026effectsofsem pages 17-17'\n\nBibTex:\n@article{timberlake2026effectsofsem,\n author = \"Timberlake, D. M. and Payton, E. J.\",\n title = \"Effects of SEM Distortions on EBSD Grain Size Measurements\",\n year = \"2026\",\n journal = \"Metallography, Microstructure, and Analysis\",\n volume = \"15\",\n pages = \"489-505\",\n month = \"Feb\",\n doi = \"10.1007/s13632-026-01317-3\",\n url = \"https://doi.org/10.1007/s13632-026-01317-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2192-9262\"\n}\n\n\nAbstract:\nAbstract\n Systematic discrepancies persist between results from grain size measurement standards, such as ASTM E 112 and ASTM E 2627, when different imaging techniques and analysis methods are used. The source of these discrepancies may be differences in imaging physics, resolution limits, grain boundary definitions, post-processing practices, or combinations thereof. Geometric distortions and drift effects in scanning electron microscope imaging are systematically modeled using synthetic microstructures to assess their impact on electron backscatter diffraction-based grain size measurement accuracy. Standard methods are compared to a lesser-known method, the Triple Point Count (TPC). Most distortion types introduce smaller biases than the expected accuracy of the methods under typical imaging conditions. TPC provides consistent performance across varying resolutions and sampling conditions. This provides further evidence that reported discrepancies are largely driven by an assumption made in the estimate for converting intercept measurements to ASTM grain size numbers, as recently reported by Evans et al.\n\nRelevant Snippet:\n\u2026 are re-gridded to match the same pixel spacing as the original synthetic EBSD data and any \u2026 are included to reduce error from poorly indexed or partially resolved grains. The mean \u2026\n\n\nUnobtainable Papers:\nWenyuan Zhao, Qingguo Ren, Zhihao Yao, Jie Zhao, He Jiang, and Jianxin Dong. Multi-scale calculation-aided composition optimization design of low-density and high-strength nickel-based superalloy. Jul 2023. URL: https://doi.org/10.1007/s11661-023-07133-6, doi:10.1007/s11661-023-07133-6.\nLaura Mill\u00e1n, Gabriel Kronberger, J. Ignacio Hidalgo, Ricardo Fern\u00e1ndez, Oscar Garnica, and Gaspar Gonz\u00e1lez-Doncel. Estimation of grain-level residual stresses in a quenched cylindrical sample of aluminum alloy aa5083 using genetic programming. Lecture Notes in Computer Science, pages 421-436, Jan 2021. URL: https://doi.org/10.1007/978-3-030-72699-7\\_27, doi:10.1007/978-3-030-72699-7\\_27.\nPierre Belamri, Henry Proudhon, Damien Texier, and David Ryckelynck. Quaternion-based vision-transformer for polycrystalline ebsd scans pre-trained on large-scale synthetic data. Materials & Design, 258:114599, Oct 2025. URL: https://doi.org/10.1016/j.matdes.2025.114599, doi:10.1016/j.matdes.2025.114599.\nJorge Stella and Michael Pohl. The role of crystal orientation and grain boundaries in the cavitation resistance of en 1.4301 austenitic stainless steel: an ebsd study. Dec 2021. URL: https://doi.org/10.1016/j.wear.2021.204040, doi:10.1016/j.wear.2021.204040.\nChase J. Smith, Xiao-Xiang Yu, Qianying Guo, Christopher R. Weinberger, and Gregory B. Thompson. Phase, hardness, and deformation slip behavior in mixed hf x ta 1-x c. Acta Materialia, 145:142-153, Feb 2018. URL: https://doi.org/10.1016/j.actamat.2017.11.038, doi:10.1016/j.actamat.2017.11.038.\nSeyyed Ezzatollah Moosavi, Cyril Cayron, Yandong Jing, Zeqin Liang, Elisa Cantergiani, Lo\u00efc Aron, Jonathan Friedli, and Roland Log\u00e9. Quantitative correlative sem\u2013ebsd analysis of precipitate\u2013particle and precipitate\u2013boundary interactions in al-mg-si alloys. Materials Characterization, May 2026. URL: https://doi.org/10.1016/j.matchar.2026.116328, doi:10.1016/j.matchar.2026.116328.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_uh7IvP3yQ85eCVKwBxCbjRkg", + "info": { + "start_ts": 619.177332258, + "end_ts": 655.107148383 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. High-strength additively manufacturable Al-Zr-Er-Ni alloys with high as-built ductility and thermal stability\nValid Text Names:\n'ge2025highstrengthadditivelymanufacturable pages 1-2', 'ge2025highstrengthadditivelymanufacturable pages 2-4', 'ge2025highstrengthadditivelymanufacturable pages 4-6', 'ge2025highstrengthadditivelymanufacturable pages 6-7', 'ge2025highstrengthadditivelymanufacturable pages 7-8', 'ge2025highstrengthadditivelymanufacturable pages 8-9', 'ge2025highstrengthadditivelymanufacturable pages 9-10', 'ge2025highstrengthadditivelymanufacturable pages 10-11', 'ge2025highstrengthadditivelymanufacturable pages 11-11'\n\nBibTex:\n@article{ge2025highstrengthadditivelymanufacturable,\n author = \"Ge, Zhaoxuan and Wei, Shaolou and Liu, Zehua and Choma, Tomasz and Zrodowski, Lukasz and Graef, Marc De and Hart, A. John and Taheri-Mousavi, S. Mohadeseh\",\n title = \"High-strength additively manufacturable Al-Zr-Er-Ni alloys with high as-built ductility and thermal stability\",\n year = \"2025\",\n journal = \"Unknown journal\",\n month = \"May\",\n doi = \"10.21203/rs.3.rs-6558867/v1\",\n url = \"https://doi.org/10.21203/rs.3.rs-6558867/v1\",\n publisher = \"Springer Science and Business Media LLC\"\n}\n\n\nAbstract:\nAbstract\n

Additively manufactured (AM) aluminum (Al) alloys with high as-built ductility offer sustainability, enhanced load-bearing performance, and resistance to residual-stress-induced failure. However, most commercial AM Al alloys have very limited ductility. In this study, we designed an AM Al-1.63Zr-2.38Er-1.34Ni (wt. %) alloy with 16% as-built ductility, while not introducing nanopowders or expensive scandium (Sc) to the formulation. We revealed that Al3Zr accounted for 65% of cavitation in the as-built condition by SEM characterizations. With calculation of phase diagram (CALPHAD)-based integrated computational materials engineering (ICME) simulations, we found that at each Er composition, Zr must not exceed the threshold required to act as a reservoir for the L12 phase transformation (needed for high strength) during aging. An optimized composition was identified following the guideline of Zr = 1.07 \u00d7 Er; this composition enhances the ductility from 0.9% to 16% (1600% increase) in the as-built sample, while only decreasing strength by 27% compared to the benchmark alloy. The strength is preserved even after aging for 48 hours at 400\u00b0C.

\n\nRelevant Snippet:\n\u2026 through approaches such as laser parameter optimization21,26\u201333, \u2026 Accurate indexing is important in EBSD quantification \u2026 on indexing fidelity, especially for traditional Hough Indexing \u2026\n\n\n---\n\n## 2. Phase Mapping in EBSD Using Convolutional Neural Networks\nValid Text Names:\n'kaufmann2020phasemappingin pages 1-2', 'kaufmann2020phasemappingin pages 2-3', 'kaufmann2020phasemappingin pages 3-4', 'kaufmann2020phasemappingin pages 4-6', 'kaufmann2020phasemappingin pages 6-7', 'kaufmann2020phasemappingin pages 7-9', 'kaufmann2020phasemappingin pages 9-11', 'kaufmann2020phasemappingin pages 9-9'\n\nBibTex:\n@article{kaufmann2020phasemappingin,\n author = \"Kaufmann, Kevin and Zhu, Chaoyi and Rosengarten, Alexander S. and Maryanovsky, Daniel and Wang, Haoren and Vecchio, Kenneth S.\",\n title = \"Phase Mapping in EBSD Using Convolutional Neural Networks\",\n year = \"2020\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"26\",\n pages = \"458-468\",\n month = \"May\",\n doi = \"10.1017/s1431927620001488\",\n url = \"https://doi.org/10.1017/s1431927620001488\",\n publisher = \"Oxford University Press (OUP)\",\n issue = \"3\",\n issn = \"1431-9276\"\n}\n\n\nAbstract:\nAbstract The emergence of commercial electron backscatter diffraction (EBSD) equipment ushered in an era of information rich maps produced by determining the orientation of user-selected crystal structures. Since then, a technological revolution has occurred in the quality, rate detection, and analysis of these diffractions patterns. The next revolution in EBSD is the ability to directly utilize the information rich diffraction patterns in a high-throughput manner. Aided by machine learning techniques, this new methodology is, as demonstrated herein, capable of accurately separating phases in a material by crystal symmetry, chemistry, and even lattice parameters with fewer human decisions. This work is the first demonstration of such capabilities and addresses many of the major challenges faced in modern EBSD. Diffraction patterns are collected from a variety of samples, and a convolutional neural network, a type of machine learning algorithm, is trained to autonomously recognize the subtle differences in the diffraction patterns and output phase maps of the material. This study offers a path to machine learning coupled phase mapping as databases of EBSD patterns encompass an increasing number of the possible space groups, chemistry changes, and lattice parameter variations.\n\nRelevant Snippet:\n\u2026 , the Hough transform has continued to be the standard in pattern indexing, even though \u2026 The inability for state-of-the-art Hough-based EBSD systems to make determinations about \u2026\n\n\nUnobtainable Papers:\nXiaoxiang Yu, Sivaji Karna, Zachary Haddock, Paul Nolan, and John Carsley. Structure-based phase identification in aluminum alloys using kikuchi band and hough-based ebsd. Materials Today Communications, 50:114476, Jan 2026. URL: https://doi.org/10.1016/j.mtcomm.2025.114476, doi:10.1016/j.mtcomm.2025.114476.\nMiu Lun Lau, Abraham Burleigh, Jeff Terry, and Min Long. Materials characterization: can artificial intelligence be used to address reproducibility challenges? Journal of Vacuum Science & Technology A, Nov 2023. URL: https://doi.org/10.1116/6.0002809, doi:10.1116/6.0002809.\nSrinivas Doddapaneni, Sathish Kumar, Sathyashankara Sharma, Gowri Shankar, Manjunath Shettar, Nitesh Kumar, Ganesha Aroor, and Syed Mansoor Ahmad. Advancements in ebsd techniques: a comprehensive review on characterization of composites and metals, sample preparation, and operational parameters. Journal of Composites Science, 9:132, Mar 2025. URL: https://doi.org/10.3390/jcs9030132, doi:10.3390/jcs9030132.\nYe Wei, Baptiste Gault, Rama Srinivas Varanasi, Dierk Raabe, Michael Herbig, and Andrew J. Breen. Machine-learning-based atom probe crystallographic analysis. Ultramicroscopy, 194:15-24, Nov 2018. URL: https://doi.org/10.1016/j.ultramic.2018.06.017, doi:10.1016/j.ultramic.2018.06.017.\nYuling Yang, Feng Jiang, M. Yang, X. M. Li, H. G. Zheng, and QM Zhang. Electron backscatter diffraction analysis of strain distribution in adiabatic shear band and its nearby area in ti\u20133al\u20135mo\u20134\u00b75v alloy. Materials Science and Technology, 28:165-170, Feb 2012. URL: https://doi.org/10.1179/1743284711y.0000000007, doi:10.1179/1743284711y.0000000007.\nAnalysis and microstructure based simulation of fatigue damage in high-performance steels\nF. Niessen, A. Burrows, and A. Bastos da Silva Fanta. A systematic comparison of on-axis and off-axis transmission kikuchi diffraction. Ultramicroscopy, 186:158-170, Mar 2018. URL: https://doi.org/10.1016/j.ultramic.2017.12.017, doi:10.1016/j.ultramic.2017.12.017.\nKaustubh Venkatraman, Meriem Ben Haj Slama, Vincent Taupin, Nabila Maloufi, and Antoine Guitton. Tuning critical resolved shear stress ratios for bcc-titanium ti21s via an automated data analysis approach. Modelling and Simulation in Materials Science and Engineering, 29:055014, Jun 2021. URL: https://doi.org/10.1088/1361-651x/abfeb0, doi:10.1088/1361-651x/abfeb0.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_PGusrK3loi2SP0YX9X3dJxyb", + "info": { + "start_ts": 619.212901675, + "end_ts": 655.08205647 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Artificial Intelligence and Machine Learning in Cold Spray Additive Manufacturing: A Systematic Literature Review\nValid Text Names:\n'afsharnia2025artificialintelligenceand pages 1-2', 'afsharnia2025artificialintelligenceand pages 7-8', 'afsharnia2025artificialintelligenceand pages 8-9', 'afsharnia2025artificialintelligenceand pages 9-10', 'afsharnia2025artificialintelligenceand pages 10-11', 'afsharnia2025artificialintelligenceand pages 11-12', 'afsharnia2025artificialintelligenceand pages 12-13', 'afsharnia2025artificialintelligenceand pages 14-14', 'afsharnia2025artificialintelligenceand pages 15-16', 'afsharnia2025artificialintelligenceand pages 16-17', 'afsharnia2025artificialintelligenceand pages 17-18', 'afsharnia2025artificialintelligenceand pages 18-20', 'afsharnia2025artificialintelligenceand pages 21-23', 'afsharnia2025artificialintelligenceand pages 26-27', 'afsharnia2025artificialintelligenceand pages 30-31', 'afsharnia2025artificialintelligenceand pages 33-34', 'afsharnia2025artificialintelligenceand pages 34-36', 'afsharnia2025artificialintelligenceand pages 36-38', 'afsharnia2025artificialintelligenceand pages 38-39', 'afsharnia2025artificialintelligenceand pages 40-42', 'afsharnia2025artificialintelligenceand pages 42-44', 'afsharnia2025artificialintelligenceand pages 44-45', 'afsharnia2025artificialintelligenceand pages 45-46', 'afsharnia2025artificialintelligenceand pages 50-52', 'afsharnia2025artificialintelligenceand pages 55-56'\n\nBibTex:\n@article{afsharnia2025artificialintelligenceand,\n author = \"Afsharnia, Habib and Butt, Javaid\",\n title = \"Artificial Intelligence and Machine Learning in Cold Spray Additive Manufacturing: A Systematic Literature Review\",\n year = \"2025\",\n journal = \"Journal of Manufacturing and Materials Processing\",\n volume = \"9\",\n pages = \"334\",\n month = \"Oct\",\n doi = \"10.3390/jmmp9100334\",\n url = \"https://doi.org/10.3390/jmmp9100334\",\n publisher = \"MDPI AG\",\n issue = \"10\",\n issn = \"2504-4494\"\n}\n\n\nAbstract:\nDue to its unique benefits over conventional subtractive manufacturing, additive manufacturing methods continue to attract interest in both academia and industry. One such method is called Cold Spray Additive Manufacturing (CSAM), a solid-state coating deposition technology to manufacture repair metallic components using a gas jet and powder particles. CSAM offers low heat input, stable phases, suitability for heat-sensitive substrates, and high deposition rates. However, persistent challenges include porosity control, geometric accuracy near edges and concavities, anisotropy, and cost sensitivities linked to gas selection and nozzle wear. Interdisciplinary research across manufacturing science, materials characterisation, robotics, control, artificial intelligence (AI), and machine learning (ML) is deployed to overcome these issues. ML supports quality prediction, inverse parameter design, in situ monitoring, and surrogate models that couple process physics with data. To demonstrate the impact of AI and ML on CSAM, this study presents a systematic literature review to identify, evaluate, and analyse published studies in this domain. The most relevant studies in the literature are analysed using keyword co-occurrence and clustering. Four themes were identified: design for CSAM, material analytics, real-time monitoring and defect analytics, and deposition and AI-enabled optimisation. Based on this synthesis, core challenges are identified as small and varied datasets, transfer and identifiability limits, and fragmented sensing. Main opportunities are outlined as physics-based surrogates, active learning, uncertainty-aware inversion, and cloud-edge control for reliable and adaptable ML use in CSAM. By systematically mapping the current landscape, this work provides a critical roadmap for researchers to target the most significant challenges and opportunities in applying AI/ML to industrialise CSAM.\n\nRelevant Snippet:\n\u2026 CSAM parameter selection as optimisation over fast forward \u2026 with a particle swarm optimiser to retrieve impact angle, particle \u2026 Electron backscatter diffraction (EBSD) adds another layer \u2026\n\n\nUnobtainable Papers:\nBin Huang, Hengrui Li, Yugang Miao, Yuyang Zhao, Ji Liu, and Bintao Wu. Curvature and central angle-based morphology measurement method for digital arc additive manufacturing. Measurement, 257:118917, Jan 2026. URL: https://doi.org/10.1016/j.measurement.2025.118917, doi:10.1016/j.measurement.2025.118917.\nTao Shen and Bo Li. An integrated computation framework for predicting mechanical performance of single-phase alloys manufactured using laser powder bed fusion: a case study of cocrfemnni high-entropy alloy. Jun 2024. URL: https://doi.org/10.1016/j.mtcomm.2024.109180, doi:10.1016/j.mtcomm.2024.109180.\nChangrong Guo, Heng Xiong, Tao Ye, Jianfeng Xu, and Long Bai. Ultrasonic defect characterization in coarse-grained materials using shade-a-tfm with subarray-based parameter optimization. IEEE Transactions on Instrumentation and Measurement, 74:1-18, Jan 2025. URL: https://doi.org/10.1109/tim.2025.3562984, doi:10.1109/tim.2025.3562984.\nBingqing Chen, Longchao Zhuo, Yixing Xie, Shuai Huang, Tianyuan Wang, Taiqi Yan, Xiufang Gong, and Yanlin Wang. Comparative study on microstructure, mechanical and high temperature oxidation resistant behaviors of slm in718 superalloy before and after heat treatment. Jul 2024. URL: https://doi.org/10.1016/j.jmrt.2024.06.163, doi:10.1016/j.jmrt.2024.06.163.\nSong Zhang, Y. C. Lin, Yun-Han Ling, Dao-Guang He, Ming-Song Chen, Miao Wan, Gui-Cheng Wu, Ning-Fu Zeng, Hui-Jie Zhang, and Majid Naseri. Modeling flow behaviors and microstructure evolution of ti55511 alloy during the double-stage hot deformation process utilizing machine learning algorithm. Metals and Materials International, Jun 2026. URL: https://doi.org/10.1007/s12540-025-01995-8, doi:10.1007/s12540-025-01995-8.\nKi-Seong Park, Sujung Son, Hyo-Sun Jang, Jiwon Park, Jin-Woong Lee, Shi Woo Lee, Kee-Sun Sohn, Hyoung Seop Kim, and Chang-Seok Oh. An expandable closed-loop framework for ductility optimization in refractory high-entropy alloys. Materials Today Advances, 29:100712, Mar 2026. URL: https://doi.org/10.1016/j.mtadv.2026.100712, doi:10.1016/j.mtadv.2026.100712.\nUday M. Basheer Al-Naib. Artificial intelligence in friction stir welding of ceramic-reinforced metal composites: a review on process optimization and property prediction. The International Journal of Advanced Manufacturing Technology, 141(3-4):1095-1112, Oct 2025. URL: https://doi.org/10.1007/s00170-025-16803-1, doi:10.1007/s00170-025-16803-1.\nEmmanuel Mart\u00ednez-Guerrero, Pedro Lagos-Eulogio, Pedro Miranda-Romagnoli, Roberto Noriega-Papaqui, and Juan Carlos Seck-Tuoh-Mora. Historical elite differential evolution based on particle swarm optimization algorithm for texture optimization with application in particle physics. Applied Sciences, 14:9110, Oct 2024. URL: https://doi.org/10.3390/app14199110, doi:10.3390/app14199110.\nR. Suresh, A. Sureshkumar, V. K. Shanmuganathan, and P. Michael Joseph Stalin. Deep generative learning for microstructure evolution prediction in multimaterial welded interfaces. Welding in the World, Mar 2026. URL: https://doi.org/10.1007/s40194-026-02410-5, doi:10.1007/s40194-026-02410-5.\nHoda Zamani, Mohammad H Nadimi-Shahraki, Shokooh Taghian, and Mahdis Banaie-Dezfouli. Enhancement of bernstain-search differential evolution algorithm to solve constrained engineering problems. Text, Jan 2020. URL: https://doi.org/10.13140/rg.2.2.16902.40004, doi:10.13140/rg.2.2.16902.40004.\nSabitabrata Bhattacharya, Suman Lata Tripathi, and Vikram Kumar Kamboj. Design of tunnel fet architectures for low power application using improved chimp optimizer algorithm. Engineering with Computers, 39:1415-1458, Nov 2023. URL: https://doi.org/10.1007/s00366-021-01530-4, doi:10.1007/s00366-021-01530-4.\nLi Hu, Shu-yong Jiang, Yan-qiu Zhang, Xiao-ming Zhu, and Dong Sun. Texture evolution and inhomogeneous deformation of polycrystalline cu based on crystal plasticity finite element method and particle swarm optimization algorithm. Journal of Central South University, 24:2747-2756, Dec 2017. URL: https://doi.org/10.1007/s11771-017-3688-1, doi:10.1007/s11771-017-3688-1.\nXue-zhao Wang, Ping Zhang, Xiaomin Jiang, Xue Chen, and Youqiang Wang. Influence of solid-solution processes on the microstructure and dynamic impact behavior of cast mg-gd-y-based alloys. Aug 2025. URL: https://doi.org/10.1016/j.vacuum.2025.114280, doi:10.1016/j.vacuum.2025.114280.\nXingrui Li and Ying Guo. Particle swarm optimization based on cubic chaotic mapping and random differential mutation. Algorithms, 19:297, Apr 2026. URL: https://doi.org/10.3390/a19040297, doi:10.3390/a19040297.\nCh. Leela Kumari, Vikram Kumar Kamboj, S. K. Bath, Suman Lata Tripathi, Megha Khatri, and Shivani Sehgal. A boosted chimp optimizer for numerical and engineering design optimization challenges. Engineering with Computers, 39:1-52, Mar 2023. URL: https://doi.org/10.1007/s00366-021-01591-5, doi:10.1007/s00366-021-01591-5.\nChaoyi Zhu, Christian Kurniawan, Marcus Ochsendorf, Dayong An, Stefan Zaefferer, and Marc De Graef. Orientation, pattern center refinement and deformation state extraction through global optimization algorithms. Ultramicroscopy, 233:113407, Mar 2022. URL: https://doi.org/10.1016/j.ultramic.2021.113407, doi:10.1016/j.ultramic.2021.113407.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_KI8Kw21UgTxchIJAACgrtWy6", + "info": { + "start_ts": 619.245570574, + "end_ts": 655.096975494 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nB. S\u00f8rensen, J. Hjelen, H W \u00c5nes, and T. Breivik. Recent features in ebsd, including new trapezoidal correction for multi-mapping. IOP Conference Series: Materials Science and Engineering, 891:012021, Aug 2020. URL: https://doi.org/10.1088/1757-899x/891/1/012021, doi:10.1088/1757-899x/891/1/012021.\nN.S. De Vincentis and D.P. Field. Factors affecting confidence index in ebsd analysis. Ultramicroscopy, 225:113269, Jun 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113269, doi:10.1016/j.ultramic.2021.113269.\nMatthew M Nowell and D Parker Mancuso. Revisiting the effects of sample preparation on ebsd performance in the era of forward model indexing. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.319, doi:10.1093/mam/ozaf048.319.\nYitian Shen, Yongsheng Zhang, Wei Li, Hong Miao, Yongzhe Wang, and Yi Zeng. Local kikuchi band detection in electron backscatter diffraction patterns for enhanced pattern indexing. Nov 2021. URL: https://doi.org/10.1111/jmi.13061, doi:10.1111/jmi.13061.\nJonathan Ligda. Improving polytype identification of silicon carbide using dictionary indexing. ArXiv, Oct 2023. URL: https://doi.org/10.21236/ad1214125, doi:10.21236/ad1214125.\nEdward L. Pang, Peter M. Larsen, and Christopher A. Schuh. Resolving pseudosymmetry in tetragonal zro2 using electron backscatter diffraction with a modified dictionary indexing approach. Journal of Applied Crystallography, 53:1060-1072, Jul 2020. URL: https://doi.org/10.1107/s160057672000864x, doi:10.1107/s160057672000864x.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_AFeXRuxf8pxFGq9DkB0yS3lq", + "info": { + "start_ts": 619.277339244, + "end_ts": 655.062107148 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.\nValid Text Names:\n'burnett2016largevolumeserial pages 1-2', 'burnett2016largevolumeserial pages 2-3', 'burnett2016largevolumeserial pages 3-4', 'burnett2016largevolumeserial pages 4-6', 'burnett2016largevolumeserial pages 6-7', 'burnett2016largevolumeserial pages 7-10', 'burnett2016largevolumeserial pages 10-10', 'burnett2016largevolumeserial pages 10-11', 'burnett2016largevolumeserial pages 11-11'\n\nBibTex:\n@article{burnett2016largevolumeserial,\n author = \"Burnett, Timothy L. and Burnett, Timothy L. and Kelley, R. and Winiarski, B. and Contreras, L. and Daly, Michael and Gholinia, A. and Burke, M. G. and Withers, P. J.\",\n title = \"Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.\",\n year = \"2016\",\n journal = \"Ultramicroscopy\",\n volume = \"161\",\n pages = \"119-129\",\n month = \"Feb\",\n doi = \"10.1016/j.ultramic.2015.11.001\",\n url = \"https://doi.org/10.1016/j.ultramic.2015.11.001\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n\n\nRelevant Snippet:\n\u2026 with a dwell time=6 \u00b5s and a single frame acquisition taking 3 \u2026 equivalent quality as determined by EBSD pattern quality from \u2026 optimisation it may be possible to push the milling currents \u2026\n\n\n---\n\n## 2. Correction of artefacts associated with large area EBSD\nValid Text Names:\n'winiarski2021correctionofartefacts pages 1-4', 'winiarski2021correctionofartefacts pages 4-6', 'winiarski2021correctionofartefacts pages 6-9', 'winiarski2021correctionofartefacts pages 9-11', 'winiarski2021correctionofartefacts pages 11-15', 'winiarski2021correctionofartefacts pages 15-17', 'winiarski2021correctionofartefacts pages 17-19', 'winiarski2021correctionofartefacts pages 19-21', 'winiarski2021correctionofartefacts pages 21-22'\n\nBibTex:\n@misc{winiarski2021correctionofartefacts,\n author = \"Winiarski, B. and Gholinia, A. and Mingard, K. and Gee, M. and Thompson, G. and Withers, P.J.\",\n title = \"Correction of artefacts associated with large area EBSD\",\n year = \"2021\",\n journal = \"Ultramicroscopy\",\n volume = \"226\",\n pages = \"113315\",\n month = \"Jul\",\n doi = \"10.1016/j.ultramic.2021.113315\",\n url = \"https://doi.org/10.1016/j.ultramic.2021.113315\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n\n\nRelevant Snippet:\n\u2026 This means EBSD data are often mapped at a dwell time per \u2026 quality of the EBSD patterns. Subsequently, SE images were \u2026 Fourthly, a verification test is run to optimise the q i spacing \u2026\n\n\nUnobtainable Papers:\nMeiyue Shao, Sriram Vijayan, Peeyush Nandwana, and Joerg R. Jinschek. The effect of beam scan strategies on microstructural variations in ti-6al-4v fabricated by electron beam powder bed fusion. Nov 2020. URL: https://doi.org/10.1016/j.matdes.2020.109165, doi:10.1016/j.matdes.2020.109165.\nGert Nolze, Maria J\u00fcrgens, J\u00fcrgen Olbricht, and Aimo Winkelmann. Improving the precision of orientation measurements from technical materials via ebsd pattern matching. Acta Materialia, 159:408-415, Oct 2018. URL: https://doi.org/10.1016/j.actamat.2018.08.028, doi:10.1016/j.actamat.2018.08.028.\n\u0160t\u011bp\u00e1n Gamanov, Poczkl\u00e1n Ladislav, Antonin Dlouhy, and Tomas Kruml. A step size dependence of kam, grod and gnd parameters calculated from ebsd. Jan 2025. URL: https://doi.org/10.2139/ssrn.4789333, doi:10.2139/ssrn.4789333.\nRobert A. Schwarzer and Jarle Hjelen. High-speed orientation microscopy with offline solving sequences of ebsd patterns. Solid State Phenomena, 160:295-300, Feb 2010. URL: https://doi.org/10.4028/www.scientific.net/ssp.160.295, doi:10.4028/www.scientific.net/ssp.160.295.\nYongjun Chen, J. Hjelen, S. Gireesh, and H. Roven. Optimization of ebsd parameters for ultra\u2010fast characterization. Journal of Microscopy, 245:111-118, Feb 2012. URL: https://doi.org/10.1111/j.1365-2818.2011.03551.x, doi:10.1111/j.1365-2818.2011.03551.x.\nStuart I. Wright, Matthew M. Nowell, Ren\u00e9 de Kloe, Patrick Camus, and Travis Rampton. Electron imaging with an ebsd detector. Ultramicroscopy, 148:132-145, Jan 2015. URL: https://doi.org/10.1016/j.ultramic.2014.10.002, doi:10.1016/j.ultramic.2014.10.002.\nPascal Thome, Micheal Madigan, Shadman Tahsin Nabil, Owen Lowerey, Francisco Medina, and Sammy Tin. Columnar to equiaxed transition in electron beam powder bed fusion (eb\u2010pbf) characterized through microstructure analytics. Advanced Engineering Materials, Mar 2026. URL: https://doi.org/10.1002/adem.202503160, doi:10.1002/adem.202503160.\nM. M. Nowell and S. I. Wright. Phase differentiation via combined ebsd and xeds. Journal of Microscopy, 213:296-305, Mar 2004. URL: https://doi.org/10.1111/j.0022-2720.2004.01299.x, doi:10.1111/j.0022-2720.2004.01299.x.\nT. Arold, P. Kroo\u00df, and T. Niendorf. Pathways to exploit the multi-spot scanning strategy in electron beam additive manufacturing for control of microstructure and defect density. Jul 2024. URL: https://doi.org/10.1016/j.jmrt.2024.06.043, doi:10.1016/j.jmrt.2024.06.043.\nE. Uta, N. Gey, P. Bocher, M. Humbert, and J. Gilgert. Texture heterogeneities in \u03b1p/\u03b1s titanium forging analysed by ebsd\u2010relation to fatigue crack propagation. Journal of Microscopy, 233:451-459, Mar 2009. URL: https://doi.org/10.1111/j.1365-2818.2009.03141.x, doi:10.1111/j.1365-2818.2009.03141.x.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_rjeQSC8BesdXWTotR8LKnLxa", + "info": { + "start_ts": 619.325013734, + "end_ts": 655.072304035 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\nValid Text Names:\n'stanek2020comparisonofsegmentation pages 1-2', 'stanek2020comparisonofsegmentation pages 2-7', 'stanek2020comparisonofsegmentation pages 7-9', 'stanek2020comparisonofsegmentation pages 9-12', 'stanek2020comparisonofsegmentation pages 12-16', 'stanek2020comparisonofsegmentation pages 16-18', 'stanek2020comparisonofsegmentation pages 18-19'\n\nBibTex:\n@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n\n\nAbstract:\nA comparison study of grain segmentation in 2D and 3D electron backscatter diffraction (EBSD) measurements of polycrystalline materials, based on a misorientation threshold, is presented. Using Laguerre tessellations with orientation marks, a simulation study that demonstrates a different sensitivity of segmentation in the case of low angle and high angle grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investigated concerning segmentation, and there are given some recommendations for the misorientation thresholds. Computational tools such as DREAM 3D software enable significant data evaluation that is completed by a five-parameter analysis of grain boundaries. K e y w o r d s: electron backscatter diffraction (EBSD), grain boundaries, misorientation, tessellation\n\nRelevant Snippet:\n\u2026 Computational tools such as DREAM 3D software enable \u2026 of grains, in a single 2D slice, on the misorientation tolerance level\u2026 number of grains in a slice with the misorientation threshold \u2026\n\n\n---\n\n## 2. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\nValid Text Names:\n'folwarczny2026accurategrainboundary pages 1-3', 'folwarczny2026accurategrainboundary pages 3-5', 'folwarczny2026accurategrainboundary pages 5-9', 'folwarczny2026accurategrainboundary pages 9-12', 'folwarczny2026accurategrainboundary pages 12-15', 'folwarczny2026accurategrainboundary pages 15-18', 'folwarczny2026accurategrainboundary pages 18-20'\n\nBibTex:\n@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n\n\nRelevant Snippet:\n\u2026 Two grain boundaries displaying the same misorientation between adjacent \u2026 normally encountered in experimental EBSD data. From the grain boundary segments, GBPDs were \u2026\n\n\nUnobtainable Papers:\nWilliam C. Lenthe, McLean P. Echlin, Jean C. Stinville, Marc De Graef, and Tresa M. Pollock. Twin related domain networks in ren\u00e9 88dt. Materials Characterization, 165:110365, Jul 2020. URL: https://doi.org/10.1016/j.matchar.2020.110365, doi:10.1016/j.matchar.2020.110365.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_x3Q5do7NWtrRZiM5mvwykX6y", + "info": { + "start_ts": 619.362055206, + "end_ts": 655.052245472 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_GpTgn8r9BIdram9mIMRTK5Lk", + "type": "function", + "function": { + "arguments": "{\"question\": \"Identify explicit evidence relevant to any of these: (1) automated optimizer-driven tuning of EBSD pattern processing/indexing parameters; (2) iterative re-indexing of stored raw patterns with CI/IQ/fit/MAD/NCC objective selecting next parameters; (3) parameter sweeps or optimization using grain-level outcomes; (4) hybrid Hough then dictionary/pattern matching on low-quality residue with automated escalation criterion; (5) autonomous EBSD post-processing or feedback to SEM acquisition parameters. For each work report exact method, automated variables/objective, and limitations/non-automated parts.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "No relevant evidence found for your question.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_GpTgn8r9BIdram9mIMRTK5Lk", + "info": { + "start_ts": 668.979169114, + "end_ts": 675.651561975 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_AqiInZ3vFforOEM02yK3ZwoA", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"Reindexing\\\" EBSD \\\"evolutionary algorithm\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_RgOFyPWBn5U8Hvx3JRcTYJut", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"dictionary indexing\\\" EBSD \\\"evolutionary algorithm\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_V0DFMvNN6qIFHJFPJYNQ3VAm", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"global optimization algorithms\\\" EBSD orientation pattern center refinement deformation state extraction\"}", + "name": "paper_search" + } + }, + { + "id": "call_ukRTrcFmrr3QCZ78ACrh54nw", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"EBSD pattern center\\\" optimization normalized cross correlation\"}", + "name": "paper_search" + } + }, + { + "id": "call_viEbTlkbaVWSj9viiSsVoQeq", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"low confidence\\\" EBSD \\\"dictionary indexing\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_sNECA4HSwXwytqITdKKFiRzx", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated parameter sweep indexing confidence index Hough settings\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_AqiInZ3vFforOEM02yK3ZwoA", + "info": { + "start_ts": 689.614977119, + "end_ts": 703.90877586 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nToby Francis, Paul F. Rottmann, Andrew T. Polonsky, Marie-Agathe Charpagne, McLean P. Echlin, Veronica Anghel, David R. Jones, George T. Gray, Marc De Graef, and Tresa M. Pollock. Multimodal 3d characterization of voids in shock-loaded tantalum: implications for ductile spallation mechanisms. Acta Materialia, 215:117057, Aug 2021. URL: https://doi.org/10.1016/j.actamat.2021.117057, doi:10.1016/j.actamat.2021.117057.\nJ.C. Stinville, M.A. Charpagne, R. Maa\u00df, H. Proudhon, W. Ludwig, P.G. Callahan, F. Wang, I.J. Beyerlein, M.P. Echlin, and T.M. Pollock. Insights into plastic localization by crystallographic slip from emerging experimental and numerical approaches. Annual Review of Materials Research, 53:275-317, Jul 2023. URL: https://doi.org/10.1146/annurev-matsci-080921-102621, doi:10.1146/annurev-matsci-080921-102621.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_RgOFyPWBn5U8Hvx3JRcTYJut", + "info": { + "start_ts": 689.651392441, + "end_ts": 719.573874281 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_V0DFMvNN6qIFHJFPJYNQ3VAm", + "info": { + "start_ts": 689.685870397, + "end_ts": 710.360182466 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nDavid Fullwood and Brent Adams. High resolution orientation imaging microscopy. ArXiv, May 2012. URL: https://doi.org/10.21236/ada586789, doi:10.21236/ada586789.\nSimon Breumier. Methodology for shot-peening induced intragranular residual stress prediction. PhD thesis, Agence Bibliographique de l\u2019Enseignement Sup\u00e9rieur, 2021. URL: https://doi.org/10.70675/9bdea3e2z19e1z4843zae96z7086104f4cb7, doi:10.70675/9bdea3e2z19e1z4843zae96z7086104f4cb7.\nMatthew J. Burch, Chris M. Fancher, Srikanth Patala, Marc De Graef, and Elizabeth C. Dickey. Mapping 180\u00b0 polar domains using electron backscatter diffraction and dynamical scattering simulations. Ultramicroscopy, 173:47-51, Feb 2017. URL: https://doi.org/10.1016/j.ultramic.2016.11.013, doi:10.1016/j.ultramic.2016.11.013.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_ukRTrcFmrr3QCZ78ACrh54nw", + "info": { + "start_ts": 689.717413937, + "end_ts": 713.509185713 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\nValid Text Names:\n'shen2019convolutionalneuralnetworkbased pages 1-2', 'shen2019convolutionalneuralnetworkbased pages 2-3', 'shen2019convolutionalneuralnetworkbased pages 3-5', 'shen2019convolutionalneuralnetworkbased pages 5-6', 'shen2019convolutionalneuralnetworkbased pages 6-8', 'shen2019convolutionalneuralnetworkbased pages 8-10', 'shen2019convolutionalneuralnetworkbased pages 10-12', 'shen2019convolutionalneuralnetworkbased pages 12-13', 'shen2019convolutionalneuralnetworkbased pages 13-14', 'shen2019convolutionalneuralnetworkbased pages 14-14'\n\nBibTex:\n@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n\n\nRelevant Snippet:\n\u2026 -art dictionary indexing (DI) method significantly improves the orientation reconstruction accuracy from highly noisy EBSD \u2026 For those with low confidence, run the full dictionary search. \u2026\n\n\nUnobtainable Papers:\nDavid P. Field, Satyam Suwas, and Stuart I. Wright. Basics of SEM-Based Techniques for Electron Backscatter Diffraction, pages 27-41. Springer Nature Singapore, Jan 2025. URL: https://doi.org/10.1007/978-981-96-5346-1\\_2, doi:10.1007/978-981-96-5346-1\\_2.\nFarangis Ram and Marc De Graef. Phase differentiation by electron backscatter diffraction using the dictionary indexing approach. Acta Materialia, 144:352-364, Feb 2018. URL: https://doi.org/10.1016/j.actamat.2017.10.069, doi:10.1016/j.actamat.2017.10.069.\nAhmed A. Tiamiyu, Tyler Lucas, Edward L. Pang, Xi Chen, James M. LeBeau, and Christopher A. Schuh. Heterogeneous microstructural evolution during hydrodynamic penetration of a high-velocity copper microparticle impacting copper. Jan 2024. URL: https://doi.org/10.1016/j.mattod.2023.11.015, doi:10.1016/j.mattod.2023.11.015.\nDevi Janani Ramesh, Janis S\u00e4lker, Sameer Aman Salman, Eduardo Vega Halleslevens, Moritz to Baben, Christina Scheu, and Jochen M. Schneider. 2d vs. 3d phase analysis of an oxidized cr2alc coating. Unknown journal, Jan 2026. URL: https://doi.org/10.2139/ssrn.6989225, doi:10.2139/ssrn.6989225.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_viEbTlkbaVWSj9viiSsVoQeq", + "info": { + "start_ts": 689.749615248, + "end_ts": 718.236611284 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nSaransh Singh. Application of forward modeling to materials characterization. Text, Jan 2018. URL: https://doi.org/10.1184/r1/6714788.v1, doi:10.1184/r1/6714788.v1.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_sNECA4HSwXwytqITdKKFiRzx", + "info": { + "start_ts": 689.780833706, + "end_ts": 719.174439903 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_gQXQyhWpcdQAtqZMTz266Aco", + "type": "function", + "function": { + "arguments": "{\"question\": \"From the retrieved papers, extract exact evidence on: Shen et al. 2019 low-confidence patterns and full dictionary search; Polonsky et al. 2019 BisQue/DREAM.3D parameter sweeps and grain segmentation outcomes; global/evolutionary optimization in EBSD pattern matching or pattern-center/orientation refinement; and whether any method iteratively tunes Hough/background/binning/neighbor settings or closes feedback to SEM acquisition. State what is and is not automated.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "No relevant evidence found for your question.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_gQXQyhWpcdQAtqZMTz266Aco", + "info": { + "start_ts": 732.745937759, + "end_ts": 740.353156312 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_4klcSpImEsXAZeSqWkaHiygh", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"shen2019convolutionalneuralnetworkbased pages 1-2\", \"shen2019convolutionalneuralnetworkbased pages 8-10\", \"shen2019convolutionalneuralnetworkbased pages 10-12\", \"polonsky2019threedimensionalanalysisand pages 1-2\", \"polonsky2019threedimensionalanalysisand pages 5-6\", \"polonsky2019threedimensionalanalysisand pages 6-8\"]}", + "name": "read_text" + } + }, + { + "id": "call_G0lw5LzH1ACLfx203yBWOksS", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"stinville2022multimodaldatasetof pages 9-12\", \"stinville2022multimodaldatasetof pages 12-13\", \"sparks20243dreconstructionof pages 9-12\", \"sparks20243dreconstructionof pages 23-24\", \"kaufmann2020phasemappingin pages 1-2\", \"kaufmann2020phasemappingin pages 6-7\"]}", + "name": "read_text" + } + }, + { + "id": "call_mH0dRIstCN2AOJxkl8ecCACI", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"burnett2016largevolumeserial pages 4-6\", \"burnett2016largevolumeserial pages 6-7\", \"winiarski2021correctionofartefacts pages 4-6\", \"winiarski2021correctionofartefacts pages 15-17\", \"stanek2020comparisonofsegmentation pages 1-2\", \"stanek2020comparisonofsegmentation pages 7-9\"]}", + "name": "read_text" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "## 1. polonsky2019threedimensionalanalysisand pages 6-8 (Context ID: pqac-00000012)\n\n the\n Keeping all other pipeline parameters constant (mini- optimal number of clusters in a dataset. The derivatives are\nmum number of two neighbors, minimum allowed defect approximated via the central difference method, shown in\nsize of 125 voxels), the misorientation tolerance was tested Fig. 5b and c. A best balance between these metrics occurs\nover a wide range of values, from 0.1 to 10\u25e6of mis- at 2.0\u25e6, making this an equitable choice of misorientation\norientation in 0.1\u25e6increments, generating a total of one- tolerance for segmentation of this dataset to avoid excess\nhundred 3D reconstructions of the steel sample described consolidation of separable grains. This results in a dataset\nin \u201cLENS-Processed 304L Stainless Steel.\u201d This represents with 899 grains, wherein 98.4% of the masked data is\nsome seventy gigabytes of generated data. To assess the included in the reconstruction.\nquality of the reconstruction, two high-level metrics were\nused, the fraction of \u201cgood\u201d data, and the total number Subvolume Creation and Analysis of Large 3D\nof grains found during the reconstruction. \u201cGood\u201d data is Volumes\ndefined by the confidence index of the EBSD data, and\ntherefore contains a constant number of voxels that does not Beyond investigation of individual pipeline parameters,\nchange between reconstructions. During the reconstruction the BisQue framework also enables coupled parameteriza-\nprocess, various cleanup steps are employed which retain tion, wherein sets of variables are linked to each other.\nvarying amounts of this initial group of voxels. DREAM.3D This allows for automated generation of random samples\nfilters such as \u201cMinimum Size,\u201d \u201cMinimum Number of of equal-sized subvolumes of 3D datasets, among otherIntegr Mater Manuf Innov (2019) 8:37\u201351 43\n\n\nFig. 4 a A single feature\nidentified from the bulk using a (a)\nmisorientation tolerance of 5.0\u25e6 222 m\nfor segmentation via CCA.\n 420 mGrain orientation is shown in\nIPF coloring with the build\ndirection taken as the reference\ndirection. b The same feature is\nsplit into three separate grains\nwhen a misorientation tolerance\nof 2.0\u25e6is used for segmentation\n\n\n\n\n\n Build\n Direction\n\n\n\n\n 111\n\n\n\n\n\n 001 101\n\n (b)\n\n\n\n\n\n Grain ID\n\n\n\n 1 2 3\n\n\napplications. The characterization of subvolumes in 3D Morphological Complexity of As-printed Additive\ndatasets has been previously applied to define microstruc- Structures\ntural and property volume elements (MVEs and PVEs) for\nvarious material properties, to inform component design Due to the complex thermal gradients and remelting\n[41, 42]. phenomena inherent to additive manufacturing methods,44 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n (a) in standard analyses of 2D cross-sections [47]. Columnar\n grains near the center of melt pool, when cut perpendicu-\n larly, can appear as fine equiaxed grains, which may mimic\n traditional weld microstructures and can lead to misinter-\n pretation of data. Misinterpretation is also highly likely\n for larger grains in additive processes. Either as a result\n of multiple remelting and solidification cycles, or from a\n single solidification event with complex thermal gradients,\n individual grains are commonly observed to possess large\n misorientation gradients. One such large grain was identi-\n fied in the reconstruction of the 304L stainless steel sample\n described in \u201cMisorientation Tolerance Parameterization\n for Feature Identification.\u201d Even while using the previously\n determined misorientation tolerance of 2\u25e6for grain segmen-\n tation, large changes in orientation are observed, in excess\n of 12\u25e6from the average orientation of the grain. Changes in\n orientation are evident via differences in inverse pole figure\n (IPF) coloring of the grain shown in Fig. 6.\n The limitations of 2D characterization of additive\n structures such as those shown in Fig. 6 are demonstrated\n by examining 2D cross-sections from the 3D data volume.\n Rate of Change per Degree\n Using the coupled parameterization ability of BisQue and\n (b)\n the \u201cCrop Geometry\u201d filter in DREAM.3D, individual\n subvolumes, each 1 voxel thick (to approximate 2D cross-\n sections), were extracted from a 3D grain along the three\n principal sample directions (Fig. 6b\u2013d). Figure 6d shows\n Z-normal slices, which are those taken with a plane\n normal along the build direction of the sample, which\n is also parallel to the TriBeam sectioning direction. X-\n normal (Fig. 6b) and Y-normal (Fig. 6c) slices are those\n containing a plane normal along either of the in-plane axes\n as defined by the TriBeam sectioning direction and the\n (c) sample geometry. These slices were then segmented using\n the same misorientation tolerance used on the full volume,\n 2\u25e6, to determine how the lack of 3D data would affect\n the characterization of such a grain. As this analysis was\n only performed on voxels associated with a large feature\n from the 3D volume, features resulting from this additional\n segmentation are subfeatures of the initial 3D grain. An\n example of a Z-normal slice characterization is shown in\n Fig. 7. Since small features comprised of fewer than 64\n voxels were already removed from the 3D volume, a modest\n minimum size requirement of four pixels was applied to\nFig. 5 a The fraction of the voxels in the confidence index mask\n these slices, effectively removing what would normally beincluded in the segmentation and the total number of grains, both as\na function of the defined grain segmentation misorientation tolerance. treated as noisy pixels in typical 2D cross-sections. The 3D\nThe rate of change per degree of misorientation tolerance for b the \n\n## 2. polonsky2019threedimensionalanalysisand pages 5-6 (Context ID: pqac-00000013)\n\nnd grains containing less than 64 voxels voxels. As with the 304L sample, a minimum CI of 0.1\nwere considered too small to be resolved and were removed. was chosen for the initial mask. Grains were segmented\nThis corresponds to 4 \u00d7 4 \u00d7 4 points across a feature and an using a misorientation tolerance of 2\u25e6using the same\nequivalent diameter of 14.9 \u03bcm, roughly five times smaller segmentation algorithm as explained in \u201cMisorientation\nthan the average grain size determined from 2D analysis Tolerance Parameterization for Feature Identification.\u201d A\ndone prior to 3D characterization. Gaps remaining from minimum size requirement of 125 voxels was applied to\nthese removed features were filled via iterative dilation of delete grains that were too small to be properly charac-\nthe surrounding grains. Indexed EBSD data was uploaded terized. This corresponds to 5\u00d75\u00d75 points across a feature\nto BisQue for analysis after data collection. and an equivalent diameter of 9.3 \u03bcm, roughly five times\n smaller than the average grain size determined from 2D\nEBM Inconel 718 analysis done prior to 3D characterization. Gaps remaining\n from these removed features were filled via iterative dilation\nElectron beam melting (EBM) is a powder-bed additive of the surrounding grains. Indexed EBSD data was uploaded\ntechnique which utilizes an electron beam as a heat source. to BisQue for analysis after data collection.\nA fine layer of powder on the order of 50\u2013100 \u03bcm is\nspread over a build plate, and the electron beam selectively\nmelts the powder. The build plate is then lowered and Results\nthe process is repeated to create complex geometries in\n3D. Unlike laser melting techniques, electron beam melting Misorientation Tolerance Parameterization\nrequires vacuum processing environment, and the electron for Feature Identi\ufb01cation\nbeam heat source can also be used to achieve preheat\ntemperatures in excess of 1000 \u25e6C. A sample of EBM In order to analyze 3D tomography data, individual fea-\nInconel 718 was fabricated with 50-\u03bcm-thick build layers tures, or grains, must be segmented from the reconstructed\nand was characterized in the TriBeam system with 1.5-\u03bcm- volume. When working with 3D EBSD data, orientation\nthick slices as described previously [37], and EBSD data information exists at every voxel in the volume. Straight-\nwas collected to characterize grain texture and morphology. forward segmentation algorithms such as connected compo-\nRoughly 25 GB of data was collected from a total of 204 nent analysis (CCA) are used to identify grains from the 3D\nslices over the course of three days, resulting in a volume data using a misorientation tolerance between adjacent vox-\nof 952.5 \u00d7 501.0 \u00d7 306.0 \u03bcm. Raw EBSD patterns were els. The misorientation tolerance is one of the most critical\nnot collected, but the Hough transforms of the patterns parameters for the segmentation and analysis of EBSD data.\n\n\nFig. 3 Mask of the 3D data from\na sample of LENS-processed 816 m\n304L stainless steel generated by 834 m\nthresholding on the Confidence\nIndex (CI) of EBSD patterns\n\n\n\n\n\n 240 m\n\n\n Build\n Direction\n\n\n\n\n\n Confidence Index\n\n\n\n\n 0.1 1.042 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\nA mask is defined that distinguishes good data from noise in Neighbors,\u201d and \u201cFill Bad Data\u201d can affect whether individ-\nthe dataset by using a threshold on the electron backscatter ual voxels are retained or released from segmented features.\npattern confidence index (CI) [36]. To perform grain seg- At the end of a reconstruction, post-processing scripts are\nmentation, a seed is selected from voxel 0 within the mask used to count the number of voxels belonging to segmented\nvolume and its orientation is compared to each of its nearest features, which will always be a subset of the initial group\nneighbors that are still within the mask (up to six neigh- of \u201cgood\u201d voxels.\nbors in a rectilinear voxelized grid). If the misorientation As can be seen in Fig. 5a, these two metrics tend to have\nbetween two adjacent voxels is below the user-defined mis- opposite trends with increasing misorientation tolerance. As\norientation tolerance, then the voxel is added to the current larger misorientation tolerances are used to segment the\ngrain. This process is repeated until no additional neighbor- data, not only are more grains merged together, decreasing\ning voxels can be added to the grain. A new seed is then the overall number of grains found, but noisier data on\nselected at the smallest voxel index not previously grouped the sample surfaces can also be connected to existing\nand the process continues until all voxels in the mask have grains during segmentation. This trend is true except for\nbeen assigned to some grain ID. the lowest misorientation tolerances, below 0.9\u25e6. At these\n It is essential that an optimal misorientation tolerance lower misorientation tolerances, the minimum grain size\nvalue is used during dataset reconstruction due to the requirement of 64 voxels eliminates many of the grains\nimpact of the misorientation tolerance on the definition found during the segmentation step, and for tolerances\nof grain structure. Two-dimensional analysis of EBSD below 0.4\u25e6, no continuous grains larger than 64 voxels\ndata typically uses a misorientation tolerance of 5\u25e6, which are found, so grain-level metrics do not exist in these\nis generally sufficient for well-annealed or recrystallized reconstructions.\nmicrostructures of various morphologies [38]. However, in An optimized reconstruction will maintain a high fraction\ndeformed or highly textured materials, as are commonly of the masked voxels with high confidence index, while\nobserved in as-built additive structures [39, 40], such also reducing unnecessary merging of grains. The merging\nlarge misorientation tolerances can inadvertently merge of grains when a large misorientation tolerance is selected\ngrains that should otherwise be segmented separately. can be seen in Fig. 4. The feature in Fig. 4a is identified\nThe additional voxel connectivity provided by 3D data as a single grain when using a misorientation tolerance\nfurther exacerbates this problem, and as there is generally of 5.0\u25e6, but becomes separated into three separate grains\nnot an accepted misorientation tolerance that applies to when using a misorientation tolerance of 2.0\u25e6(Fig. 4a).\nall 3D microstructures, these values must be chosen The further separation of this feature results in grains with\nintuitively and the resulting segmentation analyzed for more reasonable grain morphologies that reflect the additive\nefficacy. This is a time-consuming, laborious process, and process. The best misorientation tolerance value is the\nthe ultimate selection of a user-defined parameter is not lowest value at which both the change in fraction of masked\nnecessarily verifiable as the optimal choice. The use of data included and the change in the number of grains\nparameterization, enabled by the BisQue framework, allows became constant, or when the derivatives of these quantities\nfor the generation of many 3D volumes to better assess these with respect to misorientation tolerance begin to converge,\nand other important reconstruction parameters. similar to the Elbow method used for determining the\n Keeping all other pipeline parameters constant (mini- optimal number of clusters in a dataset. The derivatives are\nmum number of two neighbors, minimum allowed defect approximated via the central difference method, shown in\nsize of 125 voxels), the misorientation tolerance was tested Fig. 5b and c. A best balance between these metrics occurs\nover a wide range of values, from 0.1 to 10\u25e6of mis- at 2.0\u25e6, making this an equitable choice of misorientation\norientation in 0.1\u25e6increments, generating a total of one- tolerance for segmentation of this dataset to avoid excess\nhundred 3D reconstructions of the steel sample described consolidation of separable grains. This results in a dataset\nin \u201cLENS-Processed 304L Stainless \n\n## 3. polonsky2019threedimensionalanalysisand pages 1-2 (Context ID: pqac-00000014)\n\nIntegrating Materials and Manufacturing Innovation (2019) 8:37\u201351\nhttps://doi.org/10.1007/s40192-019-00126-7\n\n TECHNICAL ARTICLE\n\n\nThree-dimensional Analysis and Reconstruction of Additively\nManufactured Materials in the Cloud-Based BisQue Infrastructure\n\nAndrew T. Polonsky1 \u00b7 Christian A. Lang2 \u00b7 Kristian G. Kvilekval2 \u00b7 Marat I. Latypov1,3 \u00b7 McLean P. Echlin1 \u00b7\nB. S. Manjunath2 \u00b7 Tresa M. Pollock1\n\n\nReceived: 30 December 2018 / Accepted: 18 February 2019 / Published online: 22 March 2019\n\u00a9 The Minerals, Metals & Materials Society 2019\n\nAbstract\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a\ncloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline\nvariables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-\ncollected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore,\na comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using\nBisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects\nfor impact in material science are also discussed.\n\nKeywords Tomography \u00b7 3D materials science \u00b7 Microstructure analytics \u00b7 Cloud infrastructure \u00b7 Additive manufacturing \u00b7\n3D metal printing\n\n\n Introduction\n Andrew T. Polonsky\n polonsky@ucsb.edu\n Recent external pressures mandated by government grants\n Christian A. Lang requiring data be accessible to the public, and internal\n clang@ucsb.edu pressures within academic communities for data sharing\n and reproducibility have led to the emergence of data-\n Kristian G. Kvilekval\n driven platforms within the scientific community [1\u20134].\n kgk@ucsb.edu\n Requirements for these platforms include providing data\n Marat I. Latypov provenance, data sharing, and hosting services for data that\n latmarat@ucsb.edu may otherwise disappear when a lab closes or hardware\n fails [5]. Recently, data repositories have shifted from\n McLean P. Echlin\n mechlin@ucsb.edu simply siloing data to include data analytics and predictive\n capabilities. Furthermore, some of these repositories are\n B. S. Manjunath directly linked with high-performance computing in order\n manj@ucsb.edu\n to populate an ever growing database, such as with\n Tresa M. Pollock thermodynamic databases such as the Materials Project [6]\n tresap@ucsb.edu and the Open Quantum Materials Database (OQMD) [7].\n More sophisticated predictive/interactive databases, such as\n1 the aforementioned thermodynamic ones, have been slower Materials Department, University of California Santa Barbara,\n Santa Barbara, CA 93106, USA to develop with the 3D Materials Science community. This\n2 is likely due to the less standardized and more diverse data Electrical and Computer Engineering Department,\n University of California Santa Barbara, Santa Barbara, modalities, the complexity and diversity of microstructures,\n CA 93106, USA and the relative infancy of the community itself.\n3 Mechanical Engineering Department, University of California In almost all regards, using 3D data to answer scientific\n Santa Barbara, Santa Barbara, CA 93106, USA questions in Materials Science is challenging due to the38 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\nneed for dataset acquisition, reconstruction, and analysis. In 3D characterization techniques, such as X-ray-based\nReconstruction of 3D datasets is a complex process with diffraction contrast tomography (DCT), the problems of\nmany degrees of freedom. The algorithms employed, and data reconstruction and analysis are also convoluted. Coher-\ntolerances selected by the user to align slices, segment ent volumes must be reconstructed from individual 2D\nfeatures, and remove data artifacts (i.e., remove noise), images via back projection techniques [24]. The diffraction\nultimately determine the quantitative data extracted from spots from the 2D projections are generally assumed to orig-\nthe volume. Significant progress on the reconstruction and inate from a single grain, limiting the characterization of\nanalysis of electron backscatter diffraction (EBSD) data has materials with large internal strain gradients [25]. Moreover,\nbeen made by a software package DREAM.3D [8], which stereological conventions established in 2D do not neces-\ncreates a unified data structure for 3D data treatment and sarily capture the real 3D microstructures, particularly at\naggregates tools and algorithms collectively developed by extrema of property distributions [26, 27]. Given the wealth\nthe community. In DREAM.3D, this process takes the form of additional data provided by 3D characterization, new\nof a pipeline, which typically consists of well over a dozen tools aimed at exploring the parameter space offered to 3D\nsteps, and in which each step has its own set of input reconstruction are required. In the following manuscript, we\nvariables and outputs. DREAM.3D typically runs in a serial describe the use of a cloud-based platform, BisQue [28, 29],\nprocess on a workstation, with predefined input parameters. enhanced with the DREAM.3D microstructural software\nDue to the long computation times required to process analysis package for the 3D reconstruction and analysis\n3D data, which often scales proportionately as dataset size of additively manufactured microstructures characterized\nand complexity increases, it is impractical to explore the using TriBeam tomography [30].\nfull parameter space by manually running reconstructions.\nTypically, only several variables are adjusted by a user until\na qualitatively acceptable reconstruction is attained. This Methods\nprocess inherently imparts bias and typically provides little\nopportunity to justify specific reconstruction variables. The following sections describe the software infrastructure\n The bias generated during reconstruction of 3D datasets and materials investigated.\nis subsequently carried through to data analysis, which,\nfor microstructural data, is still an ill-defined process. BisQue for Cloud-Based Data Storage and Analysis\nStandardization of analysis involving new characterization\ntechniques is generally a slow process. Automated orien- All data processing was performed in the cloud with BisQue\ntation imaging analysis via EBSD first emerged in the [28, 29] using a web browser interface to interact with the\nearly 90s, but standards did not emerge for nearly two 3D datasets. BisQue has backend infrastructure support for\ndecades [9, 10]. These new standards covered basic quan- web services such as Amazon S3, IRODS, and federated\ntitative analysis, including methods for grain size measure- local file systems as well as client facing tools such as\nment, as well as guidelines governing data quality, quan- Java, Python, and Matlab. A DREAM.3D [8] plugin was\ntity, and resolution sufficient to make such measurements developed for BisQue in order to incorporate the materials\n[11]. Even in 2D EBSD analysis, such standards are rare science based 3D dataset analysis tools. The database\nand do not necessarily address additional subtleties arising architecture of BisQue is such that every processing\nfrom more complex, albeit commonly studied, microstruc- execution, dataset (input and output), and workflow\ntures, including twinned materials, nanocrystalline mate- processing pipeline receives a unique web address URL.\nrials, and deformed materials [12\u201317]. Standardization of These URLs are useful for at least two reasons: tracking\ncharacterization meth\n\n## 4. shen2019convolutionalneuralnetworkbased pages 1-2 (Context ID: pqac-00000015)\n\n Acta Materialia 170 (2019) 118e131\n\n\n Contents lists available at ScienceDirect\n\n Acta Materialia\n\n\n journal homepage: www.elsevier.com/locate/actamat\n\n\nFull length article\n\nConvolutional neural network-based method for real-time orientation\nindexing of measured electron backscatter diffraction patterns\n\nYu-Feng Shen a, b, Reeju Pokharel a, *, Thomas J. Nizolek a, Anil Kumar a, Turab Lookman a\n\na Los Alamos National Laboratory, Los Alamos, NM, 87544, USA\nb Carnegie Mellon University, Pittsburgh, PA, 15213, USA\n\n\n\na r t i c l e i n f o a b s t r a c t\n\n\nArticle history: Electron backscatter diffraction (EBSD) is the most commonly used technique for obtaining spatially\nReceived 13 September 2018 resolved microstructural information from polycrystalline materials. We have developed two convolu-\nReceived in revised form tional neural network approaches based on domain transform and transfer learning to reconstruct\n13 March 2019\n crystal orientations from electron backscatter diffraction patterns. Our models are robust to experi-\nAccepted 20 March 2019\n mentally measured image noise and index orientations as fast as the highest EBSD scanning rates. We\nAvailable online 27 March 2019\n demonstrate that the quaternion norm metric is a strong indicator for assessing the reliability of the\n reconstructions in the absence of the ground truth. We demonstrate the applicability of the current\nKeywords:\n methods on a tantalum sample.Microstructure reconstruction\nConvolutional neural network \u00a9 2019 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.\nElectron backscatter diffraction\n\n\n\n\n\n1. Introduction characterization of relatively large (0.1e10 mm) polycrystalline\n samples [6,7]. EBSD provides orientation maps at sub-grain spatial\n Spatial microstructural heterogeneities at the mesoscale resolution (~200 nm) and high crystal orientation resolution (<0.5\n(~1e100 mm), including grain boundaries, phase boundaries, dis- deg). During an EBSD measurement, an electron beam is focused\nlocations, defect structures, and chemical segregation, directly onto a spot on the sample surface and a diffraction pattern is\ncontrol macroscopic material properties and behavior. Currently, recorded. Top of the line commercially available orientation imag-\nthe lack of understanding of the complex collective response of ing microscopes (OIM) can make one such measurement in less\nmesoscale features to external stimuli poses fundamental chal- than 1 ms [8]. When optimizing for speed, the pattern quality de-\nlenges in predicting materials\u2019 microstructure evolution [1,2]. This grades, which results in reduced orientation indexing \ufb01delity with\nis critical for engineering new materials with tailored properties. conventional reconstruction methods [9]. In contrast, the state-of-\nExperimental techniques have advanced signi\ufb01cantly in the last the-art dictionary indexing (DI) method signi\ufb01cantly improves the\ntwo decades, enabling destructive and non-destructive character- orientation reconstruction accuracy from highly noisy EBSD pat-\nization of material microstructures and their evolution at high terns [9,10]. However, real-time orientation indexing using the\nspatial and temporal resolutions [3e5]. With increasing data dictionary learning method is not yet viable, unable to match scan\ncollection rates and sizes of datasets per sample reaching tens of speeds which can reach as high as 1 ms/pattern, limiting the DI\nTBs, there is considerable interest in exploring how ef\ufb01ciently approach to be an off-line method [11,12].\nanalysis methods based on machine learning can be applied to the Fast orientation indexing from noisy patterns is crucial for in-\nreconstruction process. In this work, we show how convolution situ microstructure evolution [7,9] and other quantitative studies\nneural networks (CNNs) can be used as \u201csurrogate\u201d models for that require statistically signi\ufb01cant volume measurements at high\nsolving the inverse problem of reconstructing microstructures from spatial resolution [3]. In spite of progress towards fast data collec-\ndiffraction data. tion times and computation, there is a need for new methods to\n Electron backscatter diffraction (EBSD) is the most extensively accelerate the orientation indexing from noisy EBSD patterns,\nused technique for 2D and/or 3D spatially resolved crystallographic especially if we are to make progress towards real-time in-situ\n studies and 3D microstructure characterization.\n Recently, deep convolutional networks have been utilized for a\n * Corresponding author. range of imaging techniques from automated image segmentation\n E-mail address: reeju@lanl.gov (R. Pokharel).\n\n\nhttps://doi.org/10.1016/j.actamat.2019.03.026\n1359-6454/\u00a9 2019 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131 119\n\n\n[13,14] to image reconstruction for magnetic resonance imaging high rates.\n[15]. While CNNs have demonstrated immense success for prob-\nlems related to image classi\ufb01cation [16] and detection [17], only a 1.2. Outline\nfew studies have utilized CNNs to address regression problems\n[18e20] for improving the performance of a forward model. The basic principle of EBSD (measurements and simulations) is\n Liu et al. [11] demonstrated the \ufb01rst application of convolu- presented in Section 2. A brief background on CNNs is reviewed in\ntional neural networks for EBSD orientation indexing from simu- Section 3, and their application to EBSD is presented in Section 4. In\nlated diffraction patterns. Separate CNN models were trained to Section 5, we summarize our key \ufb01ndings related to the model's\npredict individual components of the crystal orientation. Jha et al. sensitivity to image noise and experimental parameters, and\n[12] extended this approach to predict the full crystal orientation discuss the computational cost of the proposed methods.\nusing a deep CNN network, employing mean disorientation in the\nloss function to account for the non-linearity of the orientation 2. Electron backscatter diffraction\nspace. However, the implementation of complicated loss functions\nled to relatively long training times. In both the studies, the models A diffraction pattern (intensity distribution) on a detector, X \u00bc\nwere only trained and tested on simulated EBSD patterns, in X\u00f0x;y\u00de, where \u00f0x; y\u00de is pixel location, can be expressed as a function\ncontrast to real experimental diffraction patterns; therefore, the of crystal orientation represented as a quaternion q \u00bc \u00f0q0;q1;q2;q3\u00de,\neffects of measurement noise were not considered. In this work, of the form:\nwe demonstrate the ef\ufb01cacy of the neural network approach for\nexperimentally measured data with varying noise levels. We pre- X \u00bc P \u00f0f\u00f0qjM ; C ; G \u00de\u00de; (1)\nsent two convolutional neural networks based EBSD reconstruc-\ntion techniques that are applicable to experimental diffraction where M parameterizes the microscope, C denotes the crystal-\npatterns and robust towards measurement noise. In particular, we lography of the sample, G represent the geometry of the setup, and\nshow that by utilizing a simulation-trained CNN to provide an P \u00f0l\u00de represents a random Poisson sample drawn from a Poisson\ninitial estimate for re-training a small part of the network or distribution with mean l. Like many other diffraction-based mea-\ncombining it with an additional network with a small amount of surement methods, reconstructing a crystallographic orientation\nexperimental data results in high prediction accuracy for from an electron backscatter diffraction (EBSD) pattern is achieved\nmeasured data. by solving an inverse problem, i.e. \ufb01nding the inverse function\n which maps a g\n\n## 5. shen2019convolutionalneuralnetworkbased pages 8-10 (Context ID: pqac-00000016)\n\n prediction. A correction for the orienta- mental diffraction patterns of the same orientation collected at\ntion error induced by the pattern center shift across a large area various exposure times. The top row is the experimentally\nEBSD scan can be applied using the method described in Singh et al. measured patterns and the bottom row corresponds to the\n[34]. While we do not use this correction factor here, as we are FusionNet transformed patterns. The measured pattern quality\nprimarily interested in demonstrating that large reconstruction signi\ufb01cantly degrades with decreasing exposure times. For shorter\nerrors can be reduced through re-training (shown in Fig. 7), ac- exposures, the reconstructed patterns are more deformed, but\ncounting for the pattern center shift as a function of scan position many features are retained in the transformed image. Fig. 8(b)\nmay further reduce the low misorientations shown in Fig. 7(f). shows the simulated pattern corresponding to the same orientation126 Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131\n\n\n\n\n\nFig. 7. (a), (d) The misorientation map between the predicted and the ground truth crystal orientations. The red points correspond to the positions from which measured diffraction\npattern and crystal orientation pairs were used for training the U-Net and the re-trained networks. (c), (f) Misorientation distribution calculated for the 100 measured patterns,\nrandomly chosen from the right most side of the sample (unseen during training), used as a testing set for the two models. The measured diffraction patterns from the remaining Ta\nsample positions were used for orientation prediction. Low misorientation angle indicates good agreement. (b), (e) Spatial map for the deviation from unity of the quaternion norm.\nLarger deviation indicates higher prediction error. (For interpretation of the references to color in this \ufb01gure legend, the reader is referred to the Web version of this article.)\n\n\n\n\n\nFig. 8. (a) EBSD detector patterns. The \ufb01rst row shows the measured diffraction images of the same orientation collected at various exposure times. The second row shows the\nFusionNet transformed images that were used as the input for the CNN model for orientation indexing. (b) The simulated pattern corresponding to the same orientation in (a). (c)\nThe distributions of misorientations between the predicted and the ground truth crystal orientations of the 601 test dataset collected at various exposures. (d) Average misori-\nentation angles for data points with less than 10+ misorientation versus the exposure times for CNN and Hough transform analyses. The error bar is the standard deviation obtained\nfrom the misorientation distribution. Both methods exhibit a decrease in accuracy with increasing noise level. Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131 127\n\nin Fig. 8(a). Misorientation distributions between the measured and pixels (pixel size is 71.4 mm). This was expected because detector\nthe predicted orientations for the entire scanned area are shown in translation and crystal rotation have approximately the same effect\nFig. 8(c). The results demonstrate that the CNN approach can, to a on the EBSD pattern. Therefore, similar to Hough transform and\ndegree, learn to index very low-quality diffraction images. dictionary methods, the proposed CNN based approach also re-\n A comparison between the CNN and Hough transform recon- quires determination of the experimental setup parameters before\nstruction accuracy is shown in Fig. 8(d). Here, we show what per- reconstruction. Therefore, a single neural network will not provide\ncentage of points have less than 10+ misorientations with the accurate predictions for all experimental geometry parameters;\nground truth, as shown by the gray bars. Note that the CNN however, it might be possible to account for the change in instru-\nconsistently achieved a higher percentage of such points for all ment parameters via re-training the weights of the previously\nexposure times. The difference is especially large for 1 ms exposure trained CNN model.\ntimes where the CNN approach resulted with 50% of all points\nhaving < 10+ misorientations, while the Hough transform method 5.4. Computational cost\nonly had 15% of the points with < 10+ misorientation. Next, us-\ning only these < 10+ misorientation points, we calculated the Compared to a dictionary approach, one advantage of using\naverage misorientation and the standard deviation, as shown in neural networks is increased computational performance.\nblue. For this choice of points, the Hough transform method has Reducing computation time is an important consideration in\nlower average misorientation and lower variance. Overall, the CNN designing a network architecture. As we can see from Equation\napproach is more robust to noise because a larger number of points (A.1), the computation time for each pixel in the convolutional layer\nhave overall lower predicted misorientations, while the Hough is proportional to the kernel size and the number of \ufb01lters, which is\ntransform method is more accurate but only for a very small frac- related to the number of computations performed by the network.\ntion of points, suggesting that only points with low noise were Thus, reducing the number of layers and keeping the kernel size\naccurately reconstructed. and \ufb01lter numbers managable for the convolutional layers will\n Recent studies have also demonstrated CNNs\u2019 robustness to increase the performance. However, a CNNs\u2019 ability to approximate\nnoisy data [35,36]. This is not surprising because convolutional complex functions depends on the network size [41,42], and the\nlayers have been shown to be capable of learning the features from prediction power will decrease with decreasing network size.\nthe training data to recover a clean image from a noisy observation The network architectures of the CNN and U-Net used in this\n[37e39]. Moreover, CNNs can learn high level features automati- work were optimized for both computational performance and\ncally and are very robust to local noise, which makes them prediction accuracy, and we demonstrated that the models gener-\noutperform traditional methods based on handcrafted features, alized well for both unseen simulated and experimental datasets.\nwhich involve pre-determining characteristic regions or de- Using the network structure described here, the U-Net method\nscriptors of an image, utilized in many computer vision tasks took about 17 min to reconstruct the 1 1 mm2 scanned region\n[15e17,40]. (251001 data points) on the sample (i.e. 4 ms per pattern) on a\n MacBook Pro with a 2.8 GHz Intel Core i7 processor, and the\n5.3. Effects of uncertainty in experimental parameters retrained CNN took about 5 min (i.e. 1 ms per pattern). Further-\n more, there are many neural network frameworks (e.g. tensor\ufb02ow\n Another challenge always encountered in experiments is the [43]) that can utilize GPUs to further speed up the reconstruction\nuncertainty in experimental setup parameters. We conducted a without changing the code.\nparametric study to investigate the effect of instrument setup un-\ncertainty on model prediction. For that purpose, we trained our 5.5. Combining CNN with the dictionary approach\nmodel with simulated images generated from one set of detector\nparameters, and tested it on the images that were generated using Both Liu et al. [11] and Jha et al. [12] reported the dictionary\ndifferent setup parameters. The results are shown in Fig. 9. In indexing approach to be slower than the CNN method. In this work,\nFig. 9(a), we only changed the distance between the detector and we have demonstrated that the CNN-based method is relatively fast\nthe sample, keeping everything else \ufb01xed. As long as the distance and can enable real-time reconstructions during the scanning\noffset was smaller than 900 mm, the results were not strongly process. However, the CNN method is not as accurate as the dic-\nin\ufb02uenced. Fig. 9(b) and (c) show the effects of X and Y translation, tionary approach, which can reconstruct noisy data with high ac-\nwhich caused the error to shift towards the right suggesting poor curacy [10]. Further improvements may be made by combining the\nprediction. The extent of this shifting was dependent on the choice two approaches: \ufb01rst apply CNN-based regression to obtain pre-\nof the experimenta\n\n## 6. shen2019convolutionalneuralnetworkbased pages 10-12 (Context ID: pqac-00000017)\n\n we have demonstrated that the CNN-based method is relatively fast\nthe sample, keeping everything else \ufb01xed. As long as the distance and can enable real-time reconstructions during the scanning\noffset was smaller than 900 mm, the results were not strongly process. However, the CNN method is not as accurate as the dic-\nin\ufb02uenced. Fig. 9(b) and (c) show the effects of X and Y translation, tionary approach, which can reconstruct noisy data with high ac-\nwhich caused the error to shift towards the right suggesting poor curacy [10]. Further improvements may be made by combining the\nprediction. The extent of this shifting was dependent on the choice two approaches: \ufb01rst apply CNN-based regression to obtain pre-\nof the experimental setup, and in our case, it was about 4+ per 20 liminary predictions and a measure of con\ufb01dence for each location\n\n\n\n\n\nFig. 9. The misorientation distributions when the training and testing patterns are generated under different detector parameter setups. (a) The sample to detector distance is\ndifferent for training and testing data, where the offset unit is mm. (b) The pattern center X coordinate on detector are different for training and testing data. (c) The pattern center Y\ncoordinate on detector are different for training and testing data. The offset units for X and Y are given in number of pixels (pixel size is 71.4 mm).128 Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131\n\n\ngiven by the norm of the predicted quaternion. For locations with methods improved the accuracy of the experimental data recon-\nhigh con\ufb01dence reconstructions, run the pattern matching only in struction. This improvement can be understood by looking at the\nthe neighborhood (e.g. 10+) of the preliminary results in the dic- activation map of the last convolutional layer, which after max\ntionary. For those with low con\ufb01dence, run the full dictionary pooling is the input to the fully connected layer.\nsearch. Fig. 10 shows one such example from the test dataset. Four\n feature maps are shown, one for each of the four \ufb01lters in the layer.\n5.6. Regularization The \ufb01rst row is the simulation image and the \ufb01lter outputs of the\n original CNN. We can see that some features related to Kikuchi\n Regularization is important for reducing the chance of over- bands have been learned. The second shows the results of inputing\n\ufb01tting. We used L2 regularizations for our models without using experimental data into the CNN model which has been trained on\ndropout [44]. Dropout is a common regularization technique used simulations only. Clearly the original CNN will respond differently\nin most CNN classi\ufb01cation problems. It randomly forces the neuron to experimental and simulation patterns. The difference was sig-\nactivations to be zero in training, but does nothing in testing. Pre- ni\ufb01cant enough that the original CNN was unable to accurately\nvious studies have shown that dropout can improve the perfor- predict quaternions for experimental patterns.\nmance of a particular CNN-based regression model. For example, The third row shows the \ufb01lter outputs after a U-Net transformed\nKendall et. al. [45] used dropout in their model to infer camera image is fed into the original CNN; the activation maps look very\npositions from a picture taken by that camera, and saw improved similar to those shown in the \ufb01rst row, which were generated from\npredictions. We tried two dropout ratios (0:3; 0:5) in our CNN the simulated image. In the fourth row, we retrained the CNN so\nmodels, but they resulted in lower \ufb01delity predictions. However, that the weights of convolutional layers were adapted to the\nwithout dropout, we obtained validation errors that were already experimental images. We can see that the activation maps are also\nas low as the training error, as seen in Fig. 4(a). The reconstruction very similar to those in the \ufb01rst row, which supports our assump-\naccuracy (mean error and variance of error) of our model for un- tion that differences between simulation and experimental pat-\nseen data and for training data was similar, indicating that over- terns are low level features and they can be learned solely by\n\ufb01tting was not taking place. convolutional layers.\n The two approaches presented here for experimental pattern\n5.7. Microstructure reconstruction indexing are simple and robust. Similar studies have being con-\n ducted on transfer learning and domain transforms [46]. One such\n Experimental EBSD data reconstruction was performed by example is a generative adversarial networks (GAN) [47] that is\nemploying two adaptive learning networks. One of the reasons for utilized to transform images from one domain to another [48,49].\nnot training another CNN model from scratch was due to the lack of Essentially, the generative networks are learning the metric for\nlabeled experimental patterns that spanned the entire orientation measuring similarity between images produced from different mo-\nspace (fundamental zone) for a given material crystal structure. We dalities, which is similar to our approach utilizing the U-Net and\nobserved that both the domain transform and transfer learning FusionNet to transform experimental patterns to simulated patterns.\n\n\n\n\n\nFig. 10. Activation maps for the last convolutional layer. Case 1: simulated pattern as input to the original CNN. Case 2: experimental pattern as an input to the original CNN. Case 3:\nU-Net transformed pattern as an input to the original CNN. Case 4: experimental pattern as input to the re-trained CNN. Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131 129\n\n\n6. Conclusions\n\n This work demonstrates the effectiveness of two CNN ap- Xs Xs X 1 hl\u00f0i;j\u00de;c \u00bc w\u00f0m;n\u00de;c;c0al\u00f0i m;j n\u00de;c \u00fe bc;c0 (A.1)proaches for microstructure reconstruction from measured EBSD\n m\u00bc s n\u00bc s c0\npatterns. Both approaches initially trained the CNN with a large\nnumber of easily generated simulation data and then utilized a which is followed by the activation function\nrelatively small number of expensive experimentally measured\npatterns to either re-train several convolutional layers or transform\nexperimental patterns into simulated patterns. Both methods were n o\n al\u00f0i;j\u00de;c \u00bc ReLU hl\u00f0i;j\u00de;c \u00bc max 0; hl\u00f0i;j\u00de;c ;applied to a tantalum sample with a body centered cubic crystal\nstructure. Noise in the patterns was one of the major contributors\nthat impacted the model prediction signi\ufb01cantly. The proposed where hl\u00f0i;j\u00de;c and al\u00f0i;j\u00de;c are the feature map intensity and activation\nCNNs demonstrated robustness to varying levels of noise in real value at \u00f0i; j\u00de pixel coordinates, s is the kernel size and c is the\nexperimentally measured data (achieved by changing exposure\n channel index in layer l. The w\u00f0m;n\u00de;c0;c are the convolutional kernel\ntimes during data collection) and successfully reconstructed the\n weights between channel c and channel c0 of neighboring layers\norientations while matching the highest EBSD scanning rates. The\n and bc;c0 is the bias term.\nproposed quaternion norm as a measure of con\ufb01dence was\n We used a 3 3 kernel for all convolutional layers with s \u00bc 1,\ndemonstrated to be a strong indicator for assessing the reliability of\n the range of m and n was m; n2f 1; 0; 1g, and zero padding was\nthe reconstructions. This will be crucial for determining the quality\n employed. The \ufb01lter numbers (i.e. number of channels) were 4; 8; 8\nof reconstructions for real applications, when the ground truth is\n \n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013, pqac-00000014, pqac-00000015, pqac-00000016, pqac-00000017]\n\nStatus: Paper Count=25 | Relevant Papers=8 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=18 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_4klcSpImEsXAZeSqWkaHiygh", + "info": { + "start_ts": 755.581905948, + "end_ts": 760.816464142 + } + }, + { + "role": "tool", + "content": "## 1. sparks20243dreconstructionof pages 23-24 (Context ID: pqac-00000000)\n\nages and contrast-enhanced\nin the optical images, they were histogram-matched using BSE images and subsequently applied to the raw images\nScikit-Image\u2019s \u201cmatch_histograms\u201d function, with the first and binary mask data types. As the first layer of the optical\noptical image serving as the reference histogram for the later data serves as the reference location of the sample\u2019s spa-\nlayers to match against. From there, the images were spa- tial information, the optical data were centered in physical\ntially registered in series using OpenCV\u2019s implementation space around the center point of the first layer\u2019s binary mask\nof the SIFT algorithm. The default parameters were used, using SITK\u2019s \u201cCenteredTransformInitializer\u201d function. The\nand feature matching was done using OpenCV\u2019s brute force centering transform from the first optical layer was then\nmatcher, keeping the top 30% of matched features for the cal- applied to each subsequent layer to ensure that any global\nculation of the transformation matrix using \u201ccv2.estimateAff- alignment shifts due to shear or sample rotation captured\ninePartial2D.\u201d This function calculates a rigid transform, in the SIFT registration of the optical images were carried\nonly allowing for rotation and translation. After alignment, through to the BSE and EBSD registration steps.\na binary mask of the sample was automatically found using The EBSD data were then registered against the updated\na combination of Otsu thresholding, binary hole filling, and BSE images using the same SITK framework, with the\nsegmenting out the binary feature with the largest area. This B-Spline mesh size being updated to 3x3 as determined in\nbinary mask is used to mask out the mounting medium (Poly- [2]. Grayscale conversions of EBSD inverse pole figure (IPF)\nfast) and the gold cube while registering the BSE data against colormaps were registered against the contrast-enhanced\nthe optical data, as well as determining the sample center for BSE images, allowing for the nonlinear registration of grain\nlater alignment with the BSE and EBSD data. boundary information to remove the distortions from EBSD\n The same general SITK multi-resolution image registra- data collection. The IPF colormaps were generated as part of\ntion framework as outlined in [2] was used for the BSE to a Python script using ORIX, an open-source Python EBSD\noptical and EBSD to BSE registrations, with differences in analysis suite [47, 48]. ORIX\u2019s \u201corientation2color\u201d func-\nresolution and hyperparameters as needed. This framework tion was used to generate the IPF colormaps for each layer,\nallows for intra- and inter-modal image registration using while binary masks for the EBSD data were calculated using\na variety of different similarity metrics and transforma- the Otsu threshold method on the confidence index values\ntion types. The parameters used were as follows: B-Spline from the EBSD scan data. The BSE images were down-\ntransform, mutual information as the similarity metric [46], sampled from a 2500\u00d72500 grid with a spacing of 0.5 \u03bcm/\nL-BFGS-B as the optimizer, and nearest-neighbor interpola- pixel to a 600\u00d7600 grid with a spacing of 2 \u03bcm/pixel using\ntion for final image resampling. As SITK image objects are SITK\u2019s \u201cResample\u201d function to match the SITK grid used\narray-based images defined on a physical-space region, each for processing the EBSD data. Note that both the BSE and\ndata mode\u2019s image is provided along with its physical grid EBSD grids in SITK were larger than the original images\nspacing ( \u03bcm/pixel). This allows for data modes at different to allow for motion during registration without loss of data;\nresolutions to be registered, as well as the easy resampling added \u201cpadding\u201d pixels near the boundaries of the grids\nof higher-resolution data modes down to the lowest resolu- were filled with zero values to match the dark background\ntion reference grid. of both images. The B-Spline transform was then calculated\n The BSE images were registered against the aligned on a per-layer basis and applied to the raw EBSD orientation\noptical data using an updated version of the Python script data files with nearest-neighbor interpolation using a cus-\nfrom [2]. One difference from the earlier study is that only tom Python input\u2013output function. Any orientations in the\nthe BSE images and not the optical images were contrast- exported EBSD data corresponding to padding pixels were\nenhanced, as contrast enhancement of the optical images set to the identity rotation with a confidence index of zero.\nfrom this study emphasized features such as surface Separately from the SEM data modes, the SWLI data\nscratches that were not shared with the BSE images and were also registered against the self-aligned optical images.\nresulted in a poor nonlinear registration. The BSE images The raw heightmap data were corrected to remove any tilt\nwere treated with Scikit-Image\u2019s \u201cequalize_adapthist\u201d func- present from data collection, which was done by calculating\ntion to perform adaptive histogram equalization on each the surface plane of the sample using least squares estima-\nlayer, followed by a \u201cmatch_histograms\u201d treatment, match- tion and subtracting the resulting plane from the raw data.\ning each sequential layer with the first BSE layer to improve The tilt-corrected heightmaps were then registered against\ngrayscale consistency across the image series. Binary masks the optical images using the same SITK framework and\nfor the BSE images were then calculated using the same subsequently downsampled to a 600\u00d7600 grid with 2 \u03bc\u200am\nmethod as the optical images. spacing to match the grid used for the corrected EBSD data,\n For the BSE to optical registration, the B-Spline ensuring that each height value corresponded to a single\nparameters were a mesh size of 1x1 and a transformation EBSD data point. Figure 6 (d) and (h) shows an example of\norder of 2. The B-Spline transform was calculated using the raw and tilt-corrected heightmaps.796 Integrating Materials and Manufacturing Innovation (2024) 13:773\u2013803\n\n\n\n\n\nFig.\u202f26\u2002 \u2009SEM micrographs for section 331 showing the damaged region in the top left corner of the sample\n\nAppendix G.1: Registration Challenges are usually given as distances from the origin of a chosen\n reference frame, while crystal orientations are generally\nChallenges with the B-Spline registration of EBSD to BSE specified as rotations relative to an \u201corigin\u201d reference frame.\ndata occurred when one corner of the sample was accidentally It is possible for the reference frame used to define the posi-\ndamaged within the mount partway through the serial section- tion values (the spatial reference frame) to be different than\ning series. As the B-Spline transform and mutual information the reference frame used to define the rotations represent-\nsimilarity metric operate on a pixel-by-pixel basis, the SITK ing orientations (the grain reference frame), so both must\noptimizer distorts the moving image as needed to maximize be considered separately. The recorded rotations may also\nthe metric score. For the damaged region, there is no longer a differ in one or more choices required when specifying a\nclear grain structure or much shared information between the rotation in numerical form [49], such as using either the\nEBSD and BSE data types, causing the optimizer to incorrectly active or passive rotation convention. As all of these choices\ndistort the damaged region with respect to the rest of the sam- frequently vary between different data formats, cross-format\nple. In some areas of the sample, the poor distortion correc- comparisons require significant attention to these details to\ntion from the damaged region causes neighboring grains to be ensure that any differences have been accounted for.\npulled away from their true locations as well. Overall, the data We employ MTEX [37] to manipulate the EBSD and\nin the damaged region are not usable, and attempts were made NF-HEDM orientation maps, as it has native support\nto minimize its impact on the nearby regions by thresholding out for many existing orientation data formats and can be\nlow-confidence EBSD data, or potentially masking out the entire extended as necessary to accommodate new ones. The\ndamaged region from the registration optimization. Figure 26a (\u200axU\u200a, yU\u200a, zU\u200a) unified coordinate system shown in Fig. 8\nand Fig. 26b show the BSE and EBSD scans on section 331, is also the one used in MTEX, with the origin in the\nat the start of the damaged region, as well as the impact of the upper left corner, xU pointing to the righ\n\n## 2. sparks20243dreconstructionof pages 9-12 (Context ID: pqac-00000001)\n\nruns.\nagonal reconstruction grid onto the square 3D-EBSD posi-\ntion grid to allow for point-to-point quantitative compari- Reference Frame Misalignments Between Different Layers\nsons. Here, we use the coordinate system employed by the of the Same Type\n3D-EBSD position data as the unified coordinate system and\nbring the HEDM data into this coordinate system, as shown To estimate the rigid-body misalignment between two differ-\nin Fig. 8. Subsequent maps use this coordinate system. ent layers of the same type, we first calculate the misorienta-\n tions between the two layers at each (\u200axU\u200a, yU\u200a) point, similarly\n to Fig. 9. Any points with misorientation angles above the\nResults and Discussion 3\u200a\u25e6 threshold are discarded, and the remaining point-to-point\n misorientations are then averaged to produce a mean mis-\nA detailed catalog of the data is tabulated and included in orientation between the two different layers. These layer-\nthe dataset archived at the MDF. In this section, we present to-layer misorientations can be averaged further to obtain\nseveral comparison studies conducted using the 3D orienta- a mean misorientation between one layer and every other\ntion map of the VOI generated from MPSS and NF-HEDM. layer, e.g., averaging the 19 results from layer 1 of the NF-\n HEDM data versus layers 2 through 20. Averaging is done in\nLayer\u2011to\u2011Layer Orientation Comparisons quaternion form to account for all misorientation parameters\n [34], although the discussion below focuses on the misori-\nOnce the EBSD and NF-HEDM maps have been registered entation angles.\nand resampled onto the unified coordinate system, they can For the reconstructed NF-HEDM data, the misorien-\nbe easily compared across corresponding points, whether tations are uniformly very small, with all layer-to-layer\nthose are from a different data type (i.e., comparing points misorientation angles below 0.05\u25e6\u200a. As shown in Fig. 10,\nwith matching (\u200axU\u200a, yU\u200a, zU\u200a) values across data types) or a dif- higher misorientation angles are found between individual\nferent layer in the same data type (i.e., comparing points with sections of EBSD data due to the sample needing to be\nmatching (\u200axU\u200a, yU\u200a) values but different zU values within a sin- removed from the SEM for mechanical polishing and the\ngle data type). In this work, a misorientation angle threshold SEM stage needing to be moved between 0\u200a\u25e6 and 70\u25e6 tilt\nof 3\u200a\u25e6 is used to define whether a pair of points belong to the for every section. Despite the higher variation between\nsame grain in two compared datasets. Figure 9 shows that sections for the EBSD data, the mean misorientation\nthe observed misorientations between EBSD sections are angle between one section and all other sections never\ngenerally either well above or well below this threshold, with exceeds 0.3\u25e6\u200a, while the individual section-to-section mis-\nthe number of points above the threshold increasing as the orientation angles never exceed 0.5\u25e6\u200a, both low enough to782 Integrating Materials and Manufacturing Innovation (2024) 13:773\u2013803\n\n\n\n\n\nFig.\u202f9\u2002 \u2009Section-to-section misorientation maps from the distortion-corrected EBSD data\n\n\n\n\n\nFig.\u202f10\u2002 \u2009Mean misorientation angles between the distortion-corrected EBSD data for a given section and either the next section or all other sec-\ntions, and an equivalent plot for the NF-HEDM data layers\n\nbe below the noise threshold of conventional EBSD. We relative to comparisons of adjacent layers within a single\nattribute this high consistency to the combination of high- data type (see Fig. 11 versus Fig. 9). Across a subset of\nprecision EBSD, pattern center fitting for every section, the VOI cropped to contain only the sample, the fraction\nautomated sample alignment routines to ensure that the of noncorresponding points (those with a misorientation\nsample is consistently positioned during SEM imaging angle of more than 3\u200a\u25e6\u200a) was found to be 46.2%. By com-\nand EBSD scanning, and an effective clamping solution paring the IPF colormaps of both data types (as shown\nto ensure that the sample mounting on the SEM stage is in Fig. 8), we found that these points typically involved\nhighly repeatable. regions where smaller grains or twins visible in the EBSD\n data were absent from the HEDM data, with the larger\nMisorientations Between Resampled EBSD and NF\u2011HEDM grains found in the HEDM reconstruction having expanded\nData Layers to cover those regions. This result is similar to the findings\n of previous EBSD/HEDM comparison studies [1\u20134], with\nComparisons between matching layers of the resampled improving the correspondence of reconstructed HEDM\nEBSD and NF-HEDM data were similar to layer-to-layer data to the ground truth being a critical research target.\ncomparisons within a single data type, but a significantly In addition to the network of noncorresponding points\nhigher fraction of noncorresponding pixels was observed near the outlines of larger grains, Fig. 11 also shows aIntegrating Materials and Manufacturing Innovation (2024) 13:773\u2013803 783\n\n\n\n\n\nFig.\u202f11\u2002 \u2009Maps of misorientation angle and misorientation axis components versus position for the first pair of resampled 3D-EBSD and NF-\nHEDM data layers\n\n\nrelatively consistent misorientation axis and angle in the any individual layer\u2019s alignment rotation and the average\nregions that do correspond. Further investigation showed alignment rotation across all layers was 0.21\u25e6\u200a, similar in\nthat this was true for all resampled EBSD layers when com- magnitude to the misorientation between EBSD sections\npared with their matching HEDM layers. This prompted the shown in Fig. 10 (as might be expected given the high inter-\nidea of applying a per-layer uniform rotation to the EBSD nal consistency of the HEDM data).\norientation data to better align it with the HEDM orienta- The source of this misorientation between the EBSD and\ntion data, empirically determined from the average misori- HEDM orientation data has not been confirmed, but the\nentation axis and angle for corresponding points in that pair high level of consistency across the entire dataset implies\nof layers. This was highly successful at reducing the mis- that it is a systematic error. For the EBSD analysis, plau-\norientation between the EBSD and HEDM data, with 95% sible sources of systematic error include the true EBSD\nof the corresponding points after such alignment having a camera tilt angle differing from its nominal value of 10\u25e6 or\nmisorientation angle below 0.46\u25e6 across the entire volume the true sample tilt in the EBSD position differing from its\nof interest, with a mean value of only 0.24\u25e6\u200a. While these nominal value of 70\u25e6\u200a. Either of these is known to produce\nalignment rotations were calculated per layer (allowing for a fictitious rotation of the crystal orientations about xU in\nthe known variations between EBSD layers to be accounted direct proportion to the difference between the nominal\nfor), the alignment rotations themselves were also remark- angle and the true angle. However, it is unlikely that the\nably consistent from layer to layer. The largest angle between latter would produce such a consistent error across multiple784 Integrating Materials and Manufacturing Innovation (2024) 13:773\u2013803\n\n\n\n\n\nFig.\u202f12\u2002 \u2009Misorientation angle between corresponding points in the ing or Hough indexing on the same raw EBSD patterns, averaged\nresampled EBSD and HEDM data (after per-layer empirical orienta- across zU and plotted versus (\u200axU\u200a, yU\u200a) position. Note the significantly\ntion alignment rotations) when using either spherical harmonic index- different color scales for the two plots\n\nsections, as the SEM stage is moved for every section. The PC shift calibration in the Y-direction was not correctly\nSEM stage tilt axis could also be slightly misaligned rela- adjusted for the large scan area used and might provide\ntive to the EBSD camera, which is known to cause consist- better performance if this was rectified. However, even in\nent errors in the measured orientations from EBSD [35]. the central region where the PC shift effec\n\n## 3. stinville2022multimodaldatasetof pages 9-12 (Context ID: pqac-00000002)\n\nlements produced within grain surface meshes, but algorithm unaware of grain ID assignment.\n (c) Final mesh after segmenting elements according to grain IDs.\n\n\n clusters of the same material connecting at a single voxel corner) that could create singularities in the finite\n element solution. The resulting geometric model represents each grain as a region (volume) with geometric\n faces (surfaces) representing grain boundaries. Attributes attached to each region allow the user to retrieve the\n grain ID as it was defined in the originating DREAM3D15 dataset. At this stage, the face geometry still reflects\n the stair-stepped boundaries between the individual voxels, therefore a geometric based algorithm is used to\n create smooth geometric faces while preserving the overall shape of the grain boundaries. The resulting geomet-\n ric model can be tagged with meshing and analysis attributes to generate a run-ready input deck for the finite\n element solver.\n For the In718 RVE, features smaller than 50 connected voxels were removed followed by an erosion/dilation\n step using a 3x3x3 block structuring element. Many of the grains in this dataset were rather thin (often only one\n or two voxels thick). These were not processed using the erosion/dilation algorithm since it would completely\n remove them from the data. Figure 13 shows a twin grain (identified as grain ID 105 in the DREAM3D dataset)\n with a cut plane showing that the grain is at times only one voxel thick. Figure 14 shows pictures of that grain as\n it goes through the reconstruction and meshing process, from the original voxel dataset to the voxel structure\n after artifacts were removed, followed by the smoothed grain surface representation and the associated volu-\n metric mesh. Figure 15(a) shows a view of the In718 RVE geometric model that was constructed from the voxel\n dataset. The corresponding mesh shown in Fig. 15(b) was created by assigning a mesh size attribute with value\n 0.3 \u03bcm on all grain boundaries, resulting in a total of 16.7 million tetrahedral elements. A close-up view of the\n\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 9www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n\n\n\n Fig. 13\u2002 Cut plane through the grain number 105. Twin grains are as thin as one voxel at times within the\n voxelized 3D dataset.\n\n\n\n\n\n Fig. 14\u2002 Grain number 105 as it passes through the modeling process. (a) Initial voxel-level grain representation.\n (b) the voxel data after artifacts were removed. (c) geometric representation of the grain and, (d) the corresponding\n discretization.\n\n\n\n\n\n Fig. 15\u2002 Modeling procedure using Simmetrix\u2019 software suite. (a) Geometric representation of the In718 RVE\n created from the serial-sectioning voxel dataset. (a) Volumetric discretization of the In718 RVE.\n\n\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 10www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n Load step E1 E2 E3 E4\n Macroscopic strain 0.69% 0.91% 1.23% 1.83%\n Average \u03b5xx strain from HR-DIC 0.68% (1.5%) 0.87% (4.4%) 1.16% (5.7%) 1.70% (7.1%)\n\n Table 1.\u2002 Comparison between macroscopic total strain and average \u03b5xx from the HR-DIC measurements.\n The absolute difference between average \u03b5xx from HR-DIC measurements and macroscopic strain is given in\n parentheses.\n\n\n\n\n\n Fig. 16\u2002 Close-up view of the RVE representation. (a) Voxelized 3D dataset. (b) Geometric model. (c)\n Volumetric mesh.\n\n\n In718 RVE model evolution from a voxel level representation to a geometric reconstruction and an associated\n volumetric mesh is provided in Fig. 16.\n Data Records\n SEM Images.\u2003 The raw SEM images used for HR-DIC calculations are 16-bit tiff images and can be down-\n loaded from the DRYAD repository29. The nomenclature and location of the images are detailed in Fig. 2. The\n image origin is located for the pixels within the tiff images, which is set as the first top-left pixel, with the x-tiff\n (Xg) proceeding to the right, and y-tiff (Zg) proceeding down the image.\n Full-field measurements from HR-DIC.\u2003 The HR-DIC data are provided as 32-bit tiff images. The images\n can be viewed using software such as ImageJ. The values at each pixel in the images are the quantitative physical\n value of the displayed full fields given in unit length for the strain field \u03b5xx, in pixels for the in-plane slip amplitude,\n and in radians for the in-plane slip direction. We refer the reader to10,12 for more information about the physical\n meaning of these quantities. The data can be downloaded from the DRYAD repository29. Adjacent pixels in all\n maps represent a physical distance of 100.34 nm. The values in the in-plane slip amplitude maps are given in pix-\n els, where one pixel represents a shearing/displacement induced by slip of 33.45 nm. The image origin is located\n for the pixels within the tiff images, which is set as the first top-left pixel, with the x-tiff (Xg) proceeding to the\n right, and y-tiff (Zg) proceeding down the image.\n 2D Crystallographic orientation data.\u2003 The IPF maps along the loading direction X (see Fig. 4) from the raw\n EBSD measurements and after cleaning and distortion (to match the DIC full field maps) are provided in a tiff format\n in a RGB color type and labeled as IPF_Raw X.tif and IPF_DistordedDIC_X.tif, respectively. The IPF map along the\n transverse direction Zg (see Fig. 4) is also provided and labeled as IPF_Raw_Z.tif. The raw crystallographic orientation\n data (Euler angle and grain structure) are provided in the .osc and .ang format and labeled as Euler_Orientation_Raw.\n osc and Euler_Orientation_Raw.ang. The .osc and .ang format are labeled as Euler Orientation Raw.osc and Euler\n Orientation Raw.ang. The .osc format can be opened using the TSL software from EDAX. The .ang format is a ASCII\n text format where the first three columns displayed the Euler angles \u03c61, \u03a6 and \u03c62 in radians. Columns 4 and 5 display\n the coordinate Xg and Zg, respectively, for a given measurement point. It should be noted that the actual data column 5\n in the .ang file are incorrectly labeled Y. Again, the label Y in column 5 corresponds to the global direction Zg. Columns\n 7 and 8 display the image quality (IQ) and confidence index (CI) associated with the collected Kikuchi patterns.\n\n 3D Dataset - voxelized structure.\u2003 The raw Kikuchi patterns obtained from EBSD measurement are pro-\n vided in the .up2 binary file format for each EBSD slice from the TriBeam tomography measurement, these\n files can be opened using EMsoft dictionary indexing30,31, EMSphInx32 spherical pattern reindexing or with the\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 11www.nature.com/scientificdata/ www.nature.com/scientific\n\n## 4. stinville2022multimodaldatasetof pages 12-13 (Context ID: pqac-00000003)\n\n 7 and 8 display the image quality (IQ) and confidence index (CI) associated with the collected Kikuchi patterns.\n\n 3D Dataset - voxelized structure.\u2003 The raw Kikuchi patterns obtained from EBSD measurement are pro-\n vided in the .up2 binary file format for each EBSD slice from the TriBeam tomography measurement, these\n files can be opened using EMsoft dictionary indexing30,31, EMSphInx32 spherical pattern reindexing or with the\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 11www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n\n\n\n Fig. 17\u2002 Experimental validation by slip trace imaging. (a) HR-DIC strain field \u03b5xx for a reduced region of\n interest. Intense strain bands show the location of slip events. (b) The associated in-plane intensity of slip given\n in nanometers. (c) BSE image for the same region of interest. BSE imaging is an established SEM modality\n to image slip events. (d) Overlap of the HR-DIC measurement (slip intensity) with the BSE image. Perfect\n agreement is observed between BSE imaging and HR-DIC measurement.\n\n\n commercial EDAX or Oxford EBSD analysis software. The data can be downloaded from the DRYAD repos-\n itory29; however, the up2 datafiles are each 3.26 GB resulting in a large approximately 1.71 TB download. The\n reconstructed DREAM3D dataset can be downloaded independently from DRYAD29 for the interested reader\n who does not require the raw up2 EBSD patterns. The indexed data from each slice are also provided in the\n EMsoft HDF .h5 formatted files. The BSE and SEM images obtained after each slice during the TriBeam exper-\n iment are provided in 8-bit tiff images. The reconstructed 3D data with the distortion correction applied is pro-\n vided in the .dream3d file (h5 structure). The Hierarchical Data Format version 5 (h5), is an open source file\n format that supports large, complex, heterogeneous data. H5 file uses a \u201cfile directory\u201d like structure that allows\n one to organize data within the h5 file in many different structured ways, as one might do in a standard computer\n file system. The h5 format also allows for embedding of metadata making it self-describing. The associated .xdmf\n file provides the container data descriptions. Such files can be opened using the open-source software Paraview\n (https://www.paraview.org/) for visualization. For more information about the different conventions used to store\n the 3D EBSD data, we encourage the reader to consult the following references18,19.\n\n 3D Dataset - meshed structures.\u2003 The mesh generated with XtalMesh is provided in two file formats, .inp\n and .vtk. The .inp file, labelled XtalMesh.inp, is used as input into ABAQUS. The .inp file format is standard to\n ABAQUS and is an ASCII data file that consists of a series of ABAQUS keyword and data lines. In it, all nodes,\n elements, and element sets of the mesh are defined. There, the mesh is represented by ten-node tetrahedral ele-\n ments (C3D10). The .vtk file, labelled XtalMesh.vtk, is a binary Visualization Toolkit (VTK) datafile provided\n for ease of visualization and analysis of the mesh using ParaView by Kitware. There, the mesh is represented by\n four-node tetrahedral elements (C3D4) to reduce system load during visualization.\n The mesh generated with the software suite from Simmetrix (SimModeler Voxel), is provided in an Ansys33\n input deck. The input deck contains the following APDL (Ansys Parametric Design Language) files: run_Model.\n mac, contains all the calls to load the mesh, assign material properties, define node components, orient each\n grain according to the average Euler angles from the EBSD measurement and, for example, assign boundary\n conditions. The file In718RVE.cdb contains element connectivity and nodal coordinates of the entire RVE\n model, assignWPCSYS.mac creates coordinate systems using Euler angles (each coordinate system ID is linked\n to the element coordinate system defined for each grain), EulerAngles.txt contains the averaged Euler angles\n (available in the .dream3d file) for each grain in the reconstructed microstructure, In718RVE_pd.cdb assigns\n element coordinate systems (ESYS command in APDL) to each element cluster that defines a grain and creates\n node components on each side of the RVE for an easy assignment of boundary conditions, assignMatProp.mac\n assigns material properties (cubic elasticity constitutive model2) to all grains, and In718RVE_ss.cdb performs\n the finite element solution. The finite element model setup presented above was tested to make sure a solution\n can be obtained with the Ansys solver.\n\n\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 12www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n\n\n\n Fig. 18\u2002 Experimental validation of the grain boundary locations. (a) Inverse pole figure map for a reduced\n region of interest obtained from the EBSD measurement performed at the surface of the specimen before\n TriBeam serial sectioning. (b) FeatureID maps at the surface of the 3d dataset of the same region of interest\n obtained after TriBeam serial sectioning and 3D reconstruction. (c) Overlap of (a and b). Very good agreement\n is obtained between the grain morphology from conventional EBSD measurement and TriBeam serial\n sectioning.\n\n\n\n\n\n Fig. 19\u2002 Experimental validation of the crystallographic orientation. Crystallographic orientation as displayed\n in the pole figure 001 of the grain labelled 820 in Fig. 18 from conventional EBSD measurement at the surface\n of the specimen and TriBeam serial sectioning. A minimal misorientation is observed between the two\n experimental measurements.\n\n\n Technical Validation\n HR-DIC Measurements.\u2003 Initial validation of the high-resolution strain fields obtained by HR-DIC is\n achieved by directly comparing the calculated average \u03b5xx strain along the loading direction over the representa-\n tive region of interest with the measured macroscopic strain along the gauge length of the specimen. HR-DIC and\n macroscopic strains are reported in Table 1.\n Good agreement is observed between the macroscoscopic total strain and average \u03b5xx from the HR-DIC\n measurements. The differences in percent of the total deformation between the two measurements are provided\n in brackets in Table 1 for the different load steps. Further validation of the HR-DIC measurements is provided\n by direct comparison of the HR-DIC fields to high resolution BSE images and displayed in Fig. 17 for a reduced\n region of interest. BSE imaging is a common measurement technique for capturing the location and existence\n of slip events at sample surfaces. Every single slip event is captured by the HR-DIC measurement and their loca-\n tions are identical to those identified in the BSE images. The intensity of the strain concentration and in-plane\n slip obtained by the HR-DIC measurement and using the Heaviside-DIC method was previously validated10. The\n intensity of slip is obtained for eac\n\n## 5. kaufmann2020phasemappingin pages 1-2 (Context ID: pqac-00000004)\n\n Microscopy and Microanalysis (2020), 1\u201311\n\n doi:10.1017/S1431927620001488\n\n\n Original Article\n\n Phase Mapping in EBSD Using Convolutional Neural Networks\n\n\n Kevin Kaufmann1 , Chaoyi Zhu2, Alexander S. Rosengarten1, Daniel Maryanovsky3, Haoren Wang1\n and Kenneth S. Vecchio1,2*\n 1Department of NanoEngineering, UC San Diego, La Jolla, CA 92093, USA; 2Materials Science and Engineering Program, UC San Diego, La Jolla, CA 92093, USA and\n 3Department of Cognitive Science, UC San Diego, La Jolla, CA 92093, USA\n\n\n Abstract\n\n The emergence of commercial electron backscatter diffraction (EBSD) equipment ushered in an era of information rich maps produced by\n determining the orientation of user-selected crystal structures. Since then, a technological revolution has occurred in the quality, rate detec-\n tion, and analysis of these diffractions patterns. The next revolution in EBSD is the ability to directly utilize the information rich diffraction\n patterns in a high-throughput manner. Aided by machine learning techniques, this new methodology is, as demonstrated herein, capable of\n accurately separating phases in a material by crystal symmetry, chemistry, and even lattice parameters with fewer human decisions. This\n work is the first demonstration of such capabilities and addresses many of the major challenges faced in modern EBSD. Diffraction patterns\n are collected from a variety of samples, and a convolutional neural network, a type of machine learning algorithm, is trained to autono-\n mously recognize the subtle differences in the diffraction patterns and output phase maps of the material. This study offers a path to\n machine learning coupled phase mapping as databases of EBSD patterns encompass an increasing number of the possible space groups,\n chemistry changes, and lattice parameter variations.\n\n Key words: convolutional neural network, crystal structure, EBSD, electron diffraction, machine learning\n\n (Received 26 December 2019; revised 1 April 2020; accepted 21 April 2020)\n\n\n\n Introduction systems (McLaren & Reddy, 2008; Karthikeyan et al., 2013).\n Despite its shortcomings, the Hough transform has continued\n Conventional electron backscatter diffraction (EBSD) is a scan-\n to be the standard in pattern indexing, even though computing\n ning electron microscope (SEM)-based technique used to deter-\n power has increased exponentially, largely because computer algo-\n mine the three-dimensional orientation of individual grains in\n rithms have previously been incapable of Kikuchi band detection,\n crystalline materials. At present, the collected electron backscatter\n or the more challenging problem of autonomously parsing relevant\n diffraction patterns (EBSPs) are primarily utilized to construct\n information for determining chemistry and the crystal structure.\n representative maps of the microstructure, given user-defined\n The inability for state-of-the-art Hough-based EBSD systems\n phases. The resulting maps and EBSPs can be further utilized\n to make determinations about individual components of or the\n to study dislocation evolution (Zhu et al., 2018) and geometrically\n entire crystal structure of a phase has not gone unnoticed in the\n necessary dislocations (Zhu et al., 2016), among others\n scientific community. One proposed solution is a dictionary-\n (Wilkinson et al., 2006; Schwartz et al., 2009; Engler & Randle,\n based approach (Park et al., 2013; Chen et al., 2015; Singh &\n 2010; Wilkinson & Britton, 2012). These analyses are currently\n De Graef, 2017; Ram & De Graef, 2018). Libraries of selected\n achieved in commercial EBSD utilizing a Hough transform on\n materials can be selected and simulated with software, such as\n each diffraction pattern combined with a look-up table of inter-\n EMSoft (Callahan & De Graef, 2013), for the purpose of compar-\n planar angles constructed from a set of selected reflectors as spec-\n ing the experimentally collected diffraction patterns with those\n ified by the user (typically up to five phases). In the process, the\n simulated. Through this process, the most similar patterns in\n Hough transform strips most of the information about the crystal\n the dictionary determine the phase and orientation of each\n structure and chemistry from the image (Michael & Goehner,\n EBSD pattern, even for deformed or fine-grained materials\n 1999; Dingley & Wright, 2009; Li & Han, 2015; Nolze et al.,\n (Ram & De Graef, 2018). These approaches have helped to allay\n 2017). The small subset of information being utilized for differen-\n major challenges in phase differentiation, such as separating mar-\n tiating the phases often results in mis-classification of similar\n tensite from cementite (Ram & De Graef, 2018). Other types of\n crystal structures (Chen & Thomson, 2010) and orientations\n phase identification methodologies have been demonstrated,\n that produce similar diffraction maxima in different crystal\n wherein EBSD is combined with other analytical techniques,\n such as energy-dispersive X-ray spectroscopy (EDS) or wave-\n length-dispersive X-ray spectroscopy (WDS), given that the\n *Author for correspondence: Kenneth S. Vecchio, E-mail: kvecchio@eng.ucsd.edu\n chemical and structural information of the phase exists in a the- Cite this article: Kaufmann K, Zhu C, Rosengarten AS, Maryanovsky D, Wang H,\n Vecchio KS (2020) Phase Mapping in EBSD Using Convolutional Neural Networks. oretical model or crystal database (Goehner & Michael, 1996;\n Microsc Microanal. doi:10.1017/S1431927620001488 Nowell & Wright, 2004; Dingley & Wright, 2009). The problem\n\n \u00a9 Microscopy Society of America 2020\n\n\nDownloaded from https://www.cambridge.org/core. Uppsala Universitetsbibliotek, on 12 May 2020 at 08:11:30, subject to the Cambridge Core terms of use, available at\nhttps://www.cambridge.org/core/terms. https://doi.org/10.1017/S1431927620001488 2 Kevin Kaufmann et al.\n\n\n remains that the collected EBSD images are traditionally pro- is being done in other applications (Ziletti et al., 2018; Oviedo\n cessed via the Hough transform when attempting to perform et al., 2019). In well-known classes of materials, application-\n phase identification. As mentioned previously, the down sampling specific models could be developed to distinguish phases tradi-\n of this data allows for patterns from multiple crystal structures to tional Hough-based EBSD finds nearly indistinguishable, such\n be mistaken for others since the computer algorithms are capable as martensite and ferrite. Moreover, modern computing hardware\n of finding orientations which may only appear to be correct allows for real-time classification in line with recent advance-\n (Karthikeyan et al., 2013). In commercial phase ident\n\n## 6. kaufmann2020phasemappingin pages 6-7 (Context ID: pqac-00000005)\n\nmposite. (a) Electron image showing the multiphase microstructure. (b) EDS map of Fe. (c) Hough transform EBSD map with\n user-selected libraries. (d) Phase map generated by Hough transform combined with non-optimized EDS measurements. (e) Phase map generated by Hough trans-\n form combined with optimized EDS measurements as reference chemistry. (f) Phase map generated from the CNN-based model\u2019s analysis of each diffraction pat-\n tern individually. Scale bars = 500 \u03bcm. There are 10,164 total EBSPs (pixels).\n\n\n This results in two body-centered cubic phases (space group 229), Wang et al. (2019) and their observation that crystal orientation\n pure Fe and \u03b1-Fe from the Al substitution in the lattice, and two influences the diffusion rates. Otherwise, the phase map is in\n primitive cubic phases (space group 221), nearly equiatomic FeAl good agreement with the expected results, and the CNN-based\n B2 and the non-equiatomic B2 solid solution. The backscattered EBSD method is adept at separating diffraction patterns based\n electron image in Figure 4a shows the alternating layers of the on space groups and small changes in chemistry within the\n material and gives an early indication that multiple phases are same space group.\n present in the sample. Figure 4b provides a reference Fe map CNNs applied to EBSD can resolve another challenging prob-\n and the bright columns indicate where pure Fe layers remain lem that phase identification including EDS reference chemistry\n after the reaction. If the FeAl B2 and pure Fe diffraction libraries for each phase does not completely resolve. The electron image\n are selected by the user, the Hough-based EBSD is unable to dif- for a cycled thermal barrier coating containing five FCC phases\n ferentiate between the two phases, and a phase map as shown in and one rhombohedral phase is shown in Figure 5c. The top\n Figure 4c is produced. Because the diffraction maxima are so sim- layer is yttria-stabilized zirconia (YS\u2013ZrO2), followed by the ther-\n ilar between these two phases, every diffraction pattern is identi- mally grown oxide layer (Al2O3). In the middle section is a com-\n fied as both phases with equal confidence, and the software plex bond coat containing Ni, a chrome-carbide phase (Cr23C6),\n selects, by default, the second phase on the user-selected list. TaC, and a small amount of Al2O3. The bottom layer is the nickel-\n Figure 4d demonstrates a representative result from a user assign- based superalloy Hastelloy X with Cr23C6 at the grain boundaries\n ing a reference chemistry from within each phase, using point from thermal cycling. Except for the Al2O3 phase, each of these\n EDS, to each of the four phases present. If the user selects the ref- phases belongs to space group 225, meaning that their crystal\n erence chemistry for each phase, with the best of intentions and symmetries are the same, but the lattice parameters can be differ-\n accuracy, a resultant map as shown in Figure 4d can be achieved. ent. As seen in Figure 5a, the Hough transform method alone is\n While the four phases are now each present, the phase map does unable to reliably differentiate the FCC phases and indexes them\n not accurately describe what is observed in the chemistry analysis. primarily as Hastelloy X with some interspersed YS\u2013ZrO2. The\n Instead, the pure Fe and \u03b1-Fe (Fe-SS) layers are interspersed ran- phase fraction of TaC (the brightest spots in the electron\n domly (rather than layers), and the B2-SS is scattered throughout image) is much too large (based on the known chemistry) and\n the region known to be near-equiatomic FeAl B2. After consider- is incorrectly included in the substrate. The Hough transform per-\n able \u201coptimization\u201d (repeatedly changing the reference chemistry forms reasonably well at identifying the band of Al2O3, which is\n selected in point EDS for each phase), the user can eventually not surprising given its rhombohedral structure. Including refer-\n reach a phase map (as shown in Fig. 4e) that looks similar to ence chemistry for each of the phases improves the overall quality\n the known microstructure; this of course can only be achieved of the Hough-based EBSD phase map, as shown in Figure 5b. The\n because the microstructure in this case is known in advance. band of TGO is further defined and the YS\u2013ZrO2 layer contains\n By comparison, a CNN trained on a small subset of patterns much fewer erroneous pixels. However, the Hastelloy X substrate\n belonging to each phase produces the phase map seen in has large grains and twins that are being identified as TaC and the\n Figure 4f. While the phases are not as perfectly linear as in Hastelloy X is indexed as approximately 40% of the bond coat\n Figure 4e, it is likely that the merging of phases at their borders layer (center section). Furthermore, the phase fraction of TaC in\n is real to some degree according to the analysis performed by the bond coat is still much higher than what is observed to be\n\n\nDownloaded from https://www.cambridge.org/core. Uppsala Universitetsbibliotek, on 12 May 2020 at 08:11:30, subject to the Cambridge Core terms of use, available at\nhttps://www.cambridge.org/core/terms. https://doi.org/10.1017/S1431927620001488 Microscopy and Microanalysis 7\n\n\n\n\n\n Fig. 5. Phase mapping a thermal barrier coating. (a) Hough transform EBSD map with user-selected libraries. (b) Phase map generated by Hough transform com-\n bined with optimized EDS measurements as reference chemistry. (c) Electron image showing the multiphase microstructure. The Hastelloy X substrate is at the\n bottom of the image. (d) Phase map generated from the CNN-based model\u2019s analysis of each diffraction pattern individually. (e\u2013l) EDS maps of the analyzed\n region. Scale bars = 100 \u03bcm. There are 45,052 total EBSPs (pixels).\n\n\n\n present in the electron image. Figure 5d is the result of training a reference pattern is typically unknown. This method is not com-\n CNN-based model to differentiate these phases. While not free of pared herein owing to the need for a dictionary of simulated pat-\n errors, the resulting phase map is observed to be a higher-fidelity terns for the initial calibration. The microstructure of commercial\n mapping of the area shown in the electron image than traditional 430 stainless steel is displayed in Figure 6a. While not yet obvious\n Hough-based EBSD offers. Comparison with the EDS maps in from the electron image alone, the martensite regions appear\n Figures 5e\u20135l further confirms the increased plausibility of the raised compared to the ferrite. The band contrast map in\n phase fractions and the credible identification of each diffraction Figure 6b confirms both the location of the martensite as well\n pattern. The YS\u2013ZrO2 layer is nearly completely indexed as such as the overlap in pattern quality between the martensite and fer-\n with most of the errors present residing at pores and cracks in the rite. This is the same problem encountered in the previous work\n oxide. The Al2O3 is now indexed as a continuous band of ther- (Goulden et al., 2017) and explains why their pattern quality-\n mally grown oxide in good agreement with the morphology of based model was dependent on iterative phase discrimination\n the Al EDS map. The central region is almost entirely indexed and \u201clocking\u201d pixels the user deemed correctly identified before\n as Ni with interspersed TaC, Cr23C6, and Al2O3. Careful compar- re-analysis. The region depicted in Figure 6a is mapped via tradi-\n ison with the EDS maps for Ta, Cr, Al, and O reveal the chemistry tional EBSD (Fig. 6c) using ferrite and martensite libraries gener-\n to overlap well with the presence, or absence, of the respective ated based on the lattice parameter from XRD and\n phase in the CNN-based phase map. The Hastelloy X substrate well-established equations from the literature (Xiao et al., 1995)\n is well confined and interlaced with Cr23C6 particles that were (refer to the Methods section for more information). The\n not identified in Figures 5a and 5b but are clearly present in Hough-based method results in nearly all the 12,236 collected dif-\n the Cr EDS map. fraction patterns being identified as martensite. In contrast, the\n The final demonstration of the potential for CNNs to revolu- machine learning model trained on \u201cstandards\u201d for martensite\n tionize EBSD technology is perhaps the most universally \n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005]\n\nStatus: Paper Count=25 | Relevant Papers=3 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=6 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_G0lw5LzH1ACLfx203yBWOksS", + "info": { + "start_ts": 755.583425656, + "end_ts": 759.976118886 + } + }, + { + "role": "tool", + "content": "## 1. burnett2016largevolumeserial pages 4-6 (Context ID: pqac-00000006)\n\ns conducted using a Nikon Metrology 225/320 kV\nWC-11 wt% Co sample were also prepared using an FEI Nova Na- Custom Bay system. The voltage used was 80 kV and the current\nnoLab FIB using a Ga\u00fe LMIS source at 30 kV and currents of 20 nA 130 \u03bcA. 3142 projections were acquired with an exposure time of\nand 1 nA. To provide a benchmark for comparison a Gatan Ilion 1 s per projection. The entire volume was reconstructed at full\nAr\u00fe Broad Ion Beam (BIB) Precision Polishing and Etching System resolution using a FDK reconstruction algorithm associated with\n(PECS) was also used to prepare a surface of the WC-11% Co the acquisition hardware. The resolution of the reconstructed vo-\nsample by removing 250 nm of material. A three-step polishing lume was (2 \u03bcm)3.\nprocedure in 2 10 3 Pa vacuum used a range of BIB conditions: Using measurements and orientation from the X-ray CT data a\n(i) 610 mA at 10 kV for 32 min., (ii) 450 mA at 7 kV for 10 min., and site for serial sectioning was prepared using the PFIB to capture a\n\ufb01nally (iii) 350 mA at 6 kV for 10 min. The sample was rocked particular region of interest where a large pore 0.5 mm below\ncontinuously by 7 30\u00b0 at a rate of 12 oscillations/min, with the the fracture surface could be seen (see Fig. 4a). The methodology\nsample surface exposed to BIB at glancing angle. to achieve this is described in detail by Burnett et al. [45]. Using\n For all the differently prepared surfaces of the WC-11 wt% Co 30 kV and a current of 1.3 \u03bcA the top as well as the sides and the\nsample identical EBSD acquisition parameters were used to allow front of a block were prepared ready for serial sectioning. For the\ndirect comparisons. EBSD maps were collected at 20 kV, 4 nA preparation of the top of the block 470 nA and 180 nA were used in\ncurrent and 14 mm working distance. The following EBSD condi- sequence to ensure a smooth top surface. 470 nA, 180 nA and \ufb01-\ntions were used: 50 nm step size, 6 ms dwell time, averaged nally 59 nA were used to prepare the front of the block face all at\nframes 1, 4 4 binning, 12 bands, RA and Hough\u00bc50 giving an 30 kV.\nacquisition rate of 164 Hz. A block of dimensions 135 \u03bcm across and 115 \u03bcm deep of\n 3D serial sectioning was conducted on a block of WC-11 wt%Co the A508 Grade 3 sample was serial sectioned. A total of 186 slices\nhard metal 150 120 80 mm3 in size. 790 slices were acquired were recorded with each 80 nm thick slice taking 6 min 20 s T.L. Burnett et al. / Ultramicroscopy 161 (2016) 119\u2013129 123\n\n\n\n\n\nFig. 4. (a) A virtual section from the X-ray CT data for the 0.5 mm diameter core machined from below a fracture surface caused by ductile fracture of a A508 Grade\n3 pressure vessel reactor steel compact tension specimen [32], (b) SEM image showing the same region being prepared by PFIB milling ready for serial sectioning.\n\n\nwith 3 min 25 s for the image acquisition and the remaining 2 min 4 4 binning, 6 bands, RA and Hough\u00bc60 giving an acquisition\n55 s predominantly milling time and some limited stage move rate of 95 Hz.\ntime (o10 s). Serial sectioning of this sample was conducted using\n59 nA current and 30 kV acceleration voltage. It was determined 2.3. Serial sectioning procedure\nexperimentally that a beam current of 59 nA was appropriate for\nserial section milling to obtain a surface \ufb01nish of suf\ufb01cient quality Fig. 5 shows how a region of interest is typically prepared for\nto then image the microstructure using the TLD detector. A rocking serial sectioning. The volume of interest is \ufb01rst protected by de-\nmill angle of 74\u00b0 was used (see Section 2.3). SEM images were positing a protective top layer of (a Pt layer of 100 100 mm2 and\ncollected using 2 kV, 800 pA and the through lens detector (TLD), a 1 mm thick is deposited with in about 10 minutes) covering the RoI\nhorizontal \ufb01eld width of 160 \u03bcm with a resolution of 6144 4096 and extending suf\ufb01ciently far to provide a uniform region where\ngave a pixel size of 26 nm. the tracking \ufb01ducials can also be milled. A front trench is milled to\n Both of the stacks of SEM images acquired from the serial create a cross section and is accompanied by the excavation of side\nsectioning were aligned using the align slices module using a least trenches, which not only de\ufb01ne the block for serial sectioning but\nsquares algorithm in Avizo 9.0.0. The aligned images were \ufb01ltered also control re-deposition during milling. Thus a block of material\nusing the Fourier Filter (FFT Filter) which removes periodic noise is de\ufb01ned which is to be serially sectioned. These blocks for serial\nsuch as the vertical curtaining effects produced from the rocking sectioning are typically prepared with a multistep milling proce-\nmilling process as such the FFT \ufb01lter was used at two angles that dure using the PFIB at 30 kV and several beam currents, starting\nmatch the rocking mill angles. A sharpening \ufb01lter was subse- with fast bulk cutting at 1.3 mA, followed by gentler milling at\nquently applied to compensate for some blurring resulting from 180 nA to de\ufb01ne the \ufb01nal shape. In each case the maximum cur-\nthe FFT \ufb01lter. For the A508 Grade 3 data an edge preserving \ufb01lter rent that is used to give the required surface \ufb01nish needs to be\nand a median \ufb01lter were then applied to improve the de\ufb01nition of determined for each different type of material.\nthe voids, inclusions and carbides. The data was then segmented To successively reveal a series of equally spaced and parallel\nusing a semi-automated approach which combined a watershed slices, each with the exposed block face undamaged, we have\nsegmentation routine to label the large voids and a top hat seg- applied and re\ufb01ned the rocking milling process integrated within a\nmentation routine to label the carbides, both procedures are in- fully automated routine. Once a cross-section is prepared (see\ncorporated into Avizo 9.0.0 software. Fig. 5) in the region of interest rocking milling was used, alter-\n The EBSD map obtained from the A508 Grade 3 steel was ac- nating between 7\u03b1 ( 8\u00b0r\u03b1r8\u00b0) relative to the conventional\nquired using a 20 kV electron beam energy, 4 nA current and vertical, top\u2013down, milling. This procedure is illustrated in Fig. 6.\n10 mm working distance. The following EBSD conditions were To start, the sample was in the imaging position (sample tilted to\nused: 75 nm step size, 2.3 ms dwell time, averaged frames 4, 52\u00b0 and 0\u00b0 rotation) so that the ion beam is incident to the normal\n\n\n\n\n\nFig. 5. Typical preparation of a region for serial sectioning tomography data acquisition prepared using the PFIB. A multi-step preparation procedure used the PFIB at 30 kV\nand 1.3 mA, 180 nA and 59 nA to prepare the block for the serial sectioning. With the sample tilted to 52\u00b0 and with 0\u00b0 rotation the block face is presented to the electron beam\n(a) SEM image and (b) PFIB ion beam image of the same region (a vertical top down view).124 T.L. Burnett et al. / Ultramicroscopy 161 (2016) 119\u2013129\n\n\n\n\n\nFig. 6. Rocking Polish Routine explained showing a diagrammatic view inside the FIB-SEM chamber (a) shows the sample in the imaging position of 52\u00b0 tilt and 0\u00b0 rotation\n(b) rotation of 90\u00b0 and tilt at 47\u00b0 for the rocking polish step at 5\u00b0 (c) rotation of 90\u00b0 and tilt 5\u00b0 for the rocking polish step at \u00fe5\u00b0. The magenta beam represents the electron\nbeam and the cyan the ion beam.\n\n\nof the sample surface and the block face is presented to the elec- from this position the stage moved the sample back to the imaging\ntron beam for imaging of the block face (Fig. 6a). The sample was position. This sequence (Fig. 6a) b) a) c) a) b) a) c) \u2026\u2026) was re-\nthen subjected to a 90\u00b0 compucentric rotation (i.e. rotation of the peated to build up the series of sections.\nsample by 90\u00b0 around the centre of the image). The sample is then Within the automated routine ion beam tracking \ufb01ducial\ntilted (which due to the rotation is now a tilt parallel to the cross markers (see Fig. 5) are used and recognised providing the pla-\nsection length), in this example case the sample was then tilted to cement of the milling location of the cross-sectional block face\n47\u00b0 giving a \u03b1\u00bc 5\u00b0 offset (Fig. 6b). The face was milled at this slices. To improve accuracy three different images of the \ufb01ducials\nposition before returning to the imaging position. After the image are automatically recorded, one from each of the different tilt\nof the block face was recorded the sample was rotated to 90\u00b0 and positions \u03b1\u00bc 5\u00b0, 0\u00b0 and \u00fe5\u00b0. This avoid\n\n## 2. burnett2016largevolumeserial pages 6-7 (Context ID: pqac-00000007)\n\n the image). The sample is then Within the automated routine ion beam tracking \ufb01ducial\ntilted (which due to the rotation is now a tilt parallel to the cross markers (see Fig. 5) are used and recognised providing the pla-\nsection length), in this example case the sample was then tilted to cement of the milling location of the cross-sectional block face\n47\u00b0 giving a \u03b1\u00bc 5\u00b0 offset (Fig. 6b). The face was milled at this slices. To improve accuracy three different images of the \ufb01ducials\nposition before returning to the imaging position. After the image are automatically recorded, one from each of the different tilt\nof the block face was recorded the sample was rotated to 90\u00b0 and positions \u03b1\u00bc 5\u00b0, 0\u00b0 and \u00fe5\u00b0. This avoids inaccuracies induced by\nthen tilted to 57\u00b0 giving the \u03b1\u00bc \u00fe5\u00b0 offset (Fig. 6c). After milling distortion of the \ufb01ducial image when recorded from the tilt\n\n\n\n\n\nFig. 7. EBSD results acquired under the same conditions showing representative phase maps (PM), band contrast (BC) maps and raw patterns for WC -11 wt% Co for surfaces\nprepared by (a) BIB at 6 kV and 350 mA, (b) PFIB at 30 kV and 59 nA, (c) FIB at 30 kV and 20 nA and (d) FIB at 30 kV and 1 nA. The band contrast maps are presented using the\nsame greyscale for comparison. Arrows indicate Co HCP phase. T.L. Burnett et al. / Ultramicroscopy 161 (2016) 119\u2013129 125\n\npositions. The FEI Slice and ViewTM (version 1.6) software in- with 1 nA using the Ga\u00fe FIB compared to 59 nA in the Xe\u00fe PFIB.\ncorporates a rocking polish as described above and was used in The way we have assessed these results is by considering what the\nthis work. maximum usable current is in order to achieve the results required.\n This is somewhat dif\ufb01cult to describe as there are many aspects to\n be considered when it comes to determining what is an appro-\n3. Results and discussion priate level of quality. As discussed in the introduction damage\n introduced by ion beam milling has multiple aspects to it with\n3.1. Milling damage study using a WC\u2013Co hard metal sample regards to chemical and physical interactions. As mentioned in the\n introduction the reactivity of Ga\u00fe is a clear but unquanti\ufb01ed dif-\n Given the susceptibility of WC\u2013Co hard metals to Ga\u00fe FIB ferentiator compared to the chemically inert Xe\u00fe ions. We ac-\nmilling damage [4,12], this material has been used to assess the knowledge that we discuss the cumulative effect of damage and in\nPFIB milling quality. Previously we have found that Ar\u00fe BIB pol- reality there are several convoluted aspects to this but most no-\nishing introduces only low levels of damage into this system tably surface amorphization, implantation and induced lattice\n[10,11] and so it is used as a benchmark here. The electron back- strain. A far more detailed study is needed but is beyond the scope\nscatter diffraction patterns generated from the Ar\u00fe BIB prepared of this work. The results presented here thus provide a benchmark\nsurface (Fig. 7a) are very high quality with high indexing rates where we have found that we can utilise 60 higher current\n(90%) for the WC phase and high band contrast metrics for Co (1 nA vs. 59 nA) to prepare surfaces of roughly equivalent quality\n(245/255) (the metrics of the average greyscale level are given out as determined by EBSD pattern quality from band contrast and\nof a full scale of 255 for the band contrast from 8 bit EBSD maps) indexing rate recorded under identical conditions. We are also\nand WC (180/255) (see Table 2). likely to have bene\ufb01tted from an improved sputtering rate, al-\n The corresponding results for the PFIB operated at 30 kV and though this has not been not quanti\ufb01ed.\n59 nA are shown in Fig. 7b. Vertical top\u2013down milling was used to An intriguing result is that there does appear to be an in\ufb02uence\nprepare this surface to give fair comparison with the FIB results, as of the current on the level of damage introduced to the specimen\n independent from the accelerating voltage. A current of 65 nA isa result some curtaining can be seen which could be removed by\n available in the FIB but the resulting quality of the \ufb01nished surfaceusing the rocking polishing technique described in Section 2.3. The\n that means that 1 nA must be used to prepare this surface whenEBSD indexing rate is only marginally worse, at 86% for the WC\n the FIB is used. Considerable systematic work is required to mapphase, than that recorded for the low damage BIB milling (see\n this parameter space of accelerating voltage and current for bothTable 2) with good values for WC (108/255) and excellent average\n the Ga\u00fe and Xe\u00fe milling parameters. The possibility of a chemicalband contrast for Co (220/255).\n interaction of the Ga\u00fe with the WC is a clear possibility but un- It is evident from Fig. 7c that poor results are achieved when\n quanti\ufb01ed here. Although glancing angle milling has been em-using the Ga\u00fe FIB at 20 nA (30 kV). The quality of the WC EBSD\n ployed the higher current may have led to a higher dose of Ga\u00fepatterns are degraded and are virtually unindexed (indexing only\n being imparted into the sample. Another possibility is that the8%) with very poor band contrast (40/255).\n spherical aberration which leads to the Ga\u00fe LMIS spot size rapidly Fig. 7d shows that when the FIB is operated at 1 nA and 30 kV\n increasingly 45 nA [32] means that even 20 nA is ineffective for\nthe quality of the EBSD patterns is much improved (see Table 2)\n preparing a cross section of this sensitive material for EBSD ana-\nbeing only slightly inferior to the results achieved with the PFIB\n lysis. It is possible that the greater \u2018tails\u2019 on the cutting beam and a\noperated at 59 nA current with an indexing rate in the WC phase\n divergence from the ideal grazing incidence of the ions impinging\nof 76% and a band contrast average greyscale of 94/255.\n on the cross section result in a deeper penetration and more ser-\n With regard to the Co phase, the band contrast is only slightly\n ious physical damage. Further work is needed and custom aper-\nworse than for BIB and PFIB, but there is some evidence (blue\n tures should be used to avoid limitations relating to discretization\nareas) that some regions of the Co matrix transformed from the\n of the currents that can be used i.e. the present con\ufb01guration of\ncubic into the hexagonal phase ( 3% HCP). The fact that different\n the PFIB provides currents of 59, 180, 470 and 1300 nA at the high\nareas were mapped due to the different times taken to prepare the\n end.\nsurfaces mean that we cannot make a quantitative comparison but\nit appears that the extent of transformation observed (taking into\n 3.2. Serial sectioning of WC\u2013Co sample\naccount regions that have developed during sintering [4]) in the\nFIB section are greater than in the PFIB and BIB datasets. It is\n Fig. 8 shows the ability of the PFI\n\n## 3. winiarski2021correctionofartefacts pages 15-17 (Context ID: pqac-00000008)\n\nis of each section, of WC-11%wt.Co\nhard metal data collected previously [11]. We present a new methodology for SE-EBSD image\npair co-registration based on trainable Combination of Shifted Filter Responses filters and two-\ndimensional Moving Least Squares method. Our approach and it\u2019s validation relies on three\nassumptions, 1) Each of our slices are planar, 2) SE images taken on the SEM, at 0\u00b0 stage tilt,\nare distortion free and 3) the approach validation was done on the reasonable expectation of\nthe large WC grains to have planar facets, which are reported elsewhere for similar CW-Co\nsystems [32,50,58\u201361] and are also seen in etched samples [32] and the abundance of WC\ngrains with straight grain edges or boundaries in the 2D images and EBSD analysis of polished\nsurfaces. The alignment of the corrected facets with the expected crystallographic planes was\nfurther validation of the corrected 3D-EBSD data. This allowed 3D-EBSD data to be corrected\nand used for quantitative analyses and 3D visualisation. This is applied to our hard metal case\nstudy giving rise to the following conclusions:\n\u2022 The cumulative long-range distortions, in the EBSD maps compared to the reference SE\n images, were quantified as being of the order of up to 10 \uf06dm across a 200 \u00d7 200 \uf06dm area\n section. They were found to significantly affect the shapes of individual grains in 2D\n leading to significant errors in grain morphologies in 3D. We should note that the distortion\n corrections effect the shape of the grains and not the orientation.\n\u2022 Curtaining effects were observed on the serial sections and the topography was measured\n using confocal microscopy and reported in our earlier paper [11], but the confocal\n microscopy measurements were not used in the correction methods here and is a subject of\n future work. However, due to the inclined angle of incidence lead to significant distortions\n (a few pixels, up to 0.3 \uf06dm) [11] of EBSD maps given that the grains are of the order of the\n grid used for long-range correction, these distortions required short-range correction. Here\n an attempt has been made to use to correct the distortions in a 3D-EBSD dataset of a hard\n metal case, with respect to the SE images and the expectation of the 3D shapes of the hard\n metal grains from the chemical extracted method as discussed earlier in section 3.\n\u2022 3D visualisation clearly showed that the volume of reconstructed microstructure using the\n raw data was severely deformed, where surface topography for a flat facet can be as high as\n 3.3 \uf06dm.\n\u2022 Using only the long-range correction, the topography for a planar facet, of a large well-\n defined WC grain, was reduced to 2.9 \uf06dm and the normal of the planar facet lay within about\n 10\u00b0 \u2013 20\u00b0 crystallographic positioning error.\n\u2022 By applying the short-range corrections, the distortion corrections reduced the maximum\n roughness of the planar facet surface topography to 0.4 \uf06dm and the misalignment of the\n surface of the grain facets to their respective crystallographic planes were reduced to about\n 7\u00b0 \u2013 12\u00b0.\nThese types of distortion correction algorithms are required to improve the accuracy of the\nEBSD maps. In this study the accuracy was based on the roughness of the reconstructed grainfacets, where a suitable material was chosen that the large WC grains are known to have planar\nfacets. Going forwards more automated procedures need to be developed so that large area\nEBSD maps can be routinely corrected and a more robust measure of accuracy needs to be\ndeveloped. This is especially important when reconstructing 3D data because registry from\nlayer to layer is important if features are not to be introduced into the reconstructed grain\nmorphologies. The correction methods outlined could be applied for 2D EBSD maps and 3D\nEBSD maps collected at different length-scales using other SST methods, e.g. mechanical\nsectioning, SEM microtome, FIB-SEM tomography and the more recently developed\ntechniques developed for large area milling such as BIB, plasma and laser technologies\n[15,16,68]. Furthermore, smart sampling techniques may be used for measuring accurate 3D\nshapes with only the imaging data and reduce the time required for orientation data\nmeasurement.\n\u2022 While the methods described above quantify and correct for in plane distortion, work also\n needs to be done to correct for the errors caused by uneven slice thickness during 3D\n reconstruction for visualisation to account for the actual spacing and misalignment of the\n successive layers in 3D-EBSD stacks. This effect has previously been shown to be\n significant [11]. However, use of the topography information gathered by confocal\n microscopy for correction is not trivial and requires further work to incorporate the\n topography measurements into the 3D visualisation software. This requires the software to\n be able to handle non-orthogonal layers, which is not the case at the moment and is the\n subject of future work.\n\n\n5. Acknowledgements\n\nThis work was supported by the grant from the University of Manchester \u2013 EPSRC \u2013 NPL research\nprogram: 3D Assessment of Surface Integrity and Performance (EP/H020047/1). Authors acknowledge\nGATAN for lending the prototype Illion broad ion beam system. The facilities now lie within, and are\nmaintained through, the Henry Royce Institute through EPSRC grants EP/R00661X/1, EP/S019367/1,\nEP/P025021/1 and EP/P025498/1. PJW is grateful to the ERC for advanced grant funding through\nCORREL-CT Grant No. 695638.\n\n\n\nDeclaration of interests\n\n\u2612 The authors declare that they have no known competing financial interests or personal\nrelationships that could have appeared to influence the work reported in this paper.\n\n\n\n6. References\n\n\n[1] G. Nolze, Image distortions in SEM and their influences on EBSD measurements,\n Ultramicroscopy. 107 (2007) 172\u2013183. https://doi.org/10.1016/j.ultramic.2006.07.003.\n[2] S. Maraghechi, J.P.M. Hoefnagels, R.H.J. Peerlings, M.G.D. Geers, Correction of scan line\n shift artifacts in scanning electron microscopy: An extended digital image correlation\n framework, Ultramicroscopy. 187 (2018). https://doi.org/10.1016/j.ultramic.2018.01.002.\n[3] R. Flatab\u00d8, A. Coste, M.M. Greve, A systematic investigation of the charging effect in\n scanning electron microscopy for metal nanostructures on insulating substrates, J. Microsc.\n 265 (2017). https://doi.org/10.1111/jmi.12497.\n[4] L.A. Giannuzzi, F.A. Stevie, Introduction to focused ion beams: Instrumentation, theory,\n techniques and practice, 2005. https://doi.org/10.1007/b101190.\n[5] B.E. Sorensen, J. Hjelen, H.W. \u00c5nes, T. Breivik, Recent features in EBSD, including new trapezoidal correction for multi-mapping, in: IOP Conf. Ser. Mater. Sci. Eng., 2020.\n https://doi.org/10.1088/1757-899X/891/1/012021.\n[6] A.J. Schwartz, M. Kumar, B.L. Adams, D.P. Field, Electron backscatter diffraction in\n materials science, 2009. https://doi.org/10.1007/978-0-387-88136-2.\n[7] D. Rowenhorst, Removing Imaging Distortions Through Automatic Stitching of EBSD\n Mosaics, Microsc. Microanal. 19 (2013). https://doi.org/10.1017/s1431927613006193.\n[8] A.L. Pilchak, A.R. Shiveley, J.S. Tiley, D.L. Ballard, AnyStitch: A tool for combining electron\n backscatter diffraction data sets, J. Microsc. 244 (2011). https://doi.org/10.1111/j.1365-\n 2818.2011.03496.x.\n[9] A. Winkelmann, G. Nolze, G. Cios, T. Tokarski, P. Ba\u0142a, Refined calibration model for\n improving the orientation precision of electron backscatter diffraction maps, Materials (Basel).\n 13 (2020). https://doi.org/10.3390/ma13122816.\n[10] B. Winiarski, G.S. Schajer, P.J. Withers, Surface Decoration for Improving the Accuracy of\n Displacement Measurements by Digital Image Correlation in SEM, Exp. Mech. (2012).\n https://doi.org/10.1007/s11340-011-9568-y.\n[11] B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G.E. Thompson, P.J. Withers, Broad ion beam\n serial section tomography, Ultramicroscopy. (2017).\n https://doi.org/10.1016/j.ultramic.2016.10.014.\n[12] F.J. Humphreys, Grain and subgrain characterisation by electron backscatter diffraction, J.\n Mater. Sci. (2001). https://doi.org/10.1023/A:1017973432592.\n[13] T.L. Burnett, P.J. Withers, Completing the picture through correlative characterization, Nat.\n Mater. (2019). https://doi.org/10.1038/s41563-019-0402-8.\n[14] G. Spanos, D.J. Rowenhorst, A.C. Lewis, A.B. Geltmacher, Combining serial sectioning,\n EBSD analysis, and image-based finite element modeling, MRS Bull. (2008).\n https://doi.org/10.1557/mrs2008.124.\n[15] M.P. Echlin, T.L. Burnett, A.T. Polonsky, T.M. Pollock, P.J. Withers, Serial sectioning in the\n SEM for three dimensional materials science, Curr. Opin. Solid State Mater. Sci. (2020)\n 100817. https://doi.org/10.1016/j.cossms.2020.100817.\n[16] A. Gholinia, M. Curd, E. Bousser, K. Taylor,\n\n## 4. winiarski2021correctionofartefacts pages 4-6 (Context ID: pqac-00000009)\n\nlity of the data can be compromised, such as in cases where there is\npoor pattern quality, low signal to noise and signal to background ratios, multiple phases with\nlow symmetry, pseudo-symmetry and complex crystal structures and highly deformed\nmicrostructures. This means EBSD data are often mapped at a dwell time per pixel of around\none to three orders of magnitude longer than SE imaging, which results in acquisition times of\nhundreds of minutes per map.\n\n As a result large area scans can be undertaken rapidly by SE imaging, where image\ndistortions and drift can be kept to less than a pixel [10], by using a combination of suitable\nbeam dwell time, image pixel resolution and high enough magnification to minimise lens\ndistortions, where applications can be found in digital image correlation (DIC) techniques [33\u2013\n36]. However, acquiring an EBSD map can be much slower, therefore the electron beam or\nthe sample are prone to drifts away from the assumed position leading to long-range image\ndistortions during the acquisition of an EBSD map [1,37,38]. Furthermore, SE images are\nusually taken at 0\u00b0 stage tilt or incidence angle, while EBSD maps are taken at 70\u00b0 stage tilt or\nincidence angle. Therefore, the EBSD maps have 1/cosine(70\u00b0) or 2.9 times smaller beam\nspacing or deflection in the direction perpendicular to the tilt axis in order to generate an\nequidistant grid of measurements points on the sample surface. This results in almost a\nthreefold degradation in spatial resolution and increase of any effect from drift or \u2018long-range\ndistortion\u2019, furthermore it accentuates the distortion from topographic features or \u2018short-range\ndistortion\u2019, in the EBSD maps [12]. Thus, these distortion exaggerations are minimised during\nSE imaging at 0\u00b0 incidence angle. This is also one of the advantages of the relatively new\ntechnique of transmission EBSD (t-EBSD) or transmission Kikuchi diffraction (TKD), where\nthe sample is usually back tilted only between 0\u00b0-20\u00b0 [39\u201342]. New approaches to EBSD\nanalysis, such as the \u2018Dictionary based indexing\u2019 [43,44] or smart acquisition [45] promise\nfaster acquisition times without compromising EBSD indexing rates, because either lower\nquality EBSD patterns and faster EBSD mapping speeds can be used or already utilizes other\nSEM imaging modes taken at higher speeds and requires less sampling points. Nevertheless,\nEBSD analysis will continue to be much slower than conventional image acquisition and thus\nmore susceptible to drifts.\nIn a previous paper introducing serial section broad ion beam tomography we identified\ndistortions in large area 2D EBSD maps [11]. It was shown that these can seriously affect the\ninterpretation of 3D EBSD datasets if uncorrected through an exemplar study of the 3D\nmicrostructure of WC-Co hard metals. In this follow-up paper we return to this problem to\nexamine the nature and to quantify the magnitude of the long-range and short-range distortions\nassociated with acquiring large area 2D EBSD data. All imaging modes in the SEM can have\ndistortions, however to correct for the large distortions encountered in the EBSD maps we\nassume that SE images are distortion free compared to the EBSD maps. Furthermore, we\npropose methods for correcting both of long and short-range distortions by comparing EBSD\nmaps to SE images. Our proposed methods utilise COSFIRE, Combination of Shifted Filter\nResponses filters [46] and Moving Least Squares (MLS) rigid deformation method for digital\nimage morphing [47]. The aim of this paper is to raise awareness of these artefacts which are\nlargely neglected at present, to quantify them and to show the benefits of their correction. This\npaper is not aimed at proposing the best method of correction, nor comparing the merits of\nspecific correction approaches, but rather to encourage a greater focus on developing such\nmethods going forwards.\n By way of a case study, this paper focuses on distortions associated with collecting\nrelatively large volume serial section data on WC-Co hard metals (or cermets). They are used\nfor machining, cutting, mining and drilling and are challenging because they contain a high\nfraction of a ceramic phase (i.e. the WC grains) held together a ductile metallic phase (i.e. theCo). The mechanical properties of cemented carbides strongly depend on the size and\nmorphology of the WC grains [48,49]. Abnormal grain growth (AGG) can have a detrimental\neffect on the mechanical properties of cermets, since these oversized grains (typically > 10 -\n30 \u00b5m) act as nucleation points for cracking, breakage and drill bit failure. The mechanism for\nAGG is not yet sufficiently understood. Some attempts have been made to simulate AGG in\ncemented carbide verified with experimental work and 2D analyses [50], but the plethora of\ninformation that can be extracted from 3D analyses allows the opportunity to much more fully\nstudy the AGG phenomenon. It must be noted here that there are non-destructive techniques\navailable to study AGG during in-situ heating [51,52], however the resolution of these\ntechniques are presently about 1 \uf06dm. Although FIB-SEM based techniques are destructive,\nbut are able to resolve volumes in 3D in nanometer resolution. Large area plasma FIB [25],\nBIB-SEM [16], laser-PFIB-SEM [24] or PFIB spin milling methods [27] are needed to capture\ndata over microstructurally relevant large volumes. As a result, such EBSD data provides a\ngood case for quantification and correction of long and short-range corrections described in\nthis paper.\n\n\n\n\n\n2. Experimental procedure and SE-EBSD image pair co-registration\n\n As mentioned above, the focus of the current paper is on the quantification of long and\nshort-range distortions associated with large area EBSD mapping and methods for their\ncorrection rather than on making serial sectioning experiments per se. To this end we return to\nthe 3D EBSD dataset collected by serial section BIB for the WC-Co hardmetal sample\ndescribed in [11].\n\n Sample\n\n The material studied was a Tungsten carbide and Cobalt (WC-Co) hardmetal containing\n11 mass % Co and having the average grain diameter of 5 \uf06dm [53]. The alloy was produced by\nMarshalls Hard Metals Ltd using conventional powder metallurgical methods. Tungsten\ncarbide (WC) has a hexagonal crystal structure with lattice parameters a=2.9\u00c5 and c=2.83\u00c5.\nThe stable Cobalt (Co) phase has a cubic crystal structure with a lattice parameter size a=3.56\u00c5.\n\n\n\n\n Broad ion beam sectioning\n The experimental BIB serial sectioning workflow followed in this study is described in\ndetail in a companion paper [11]. In essence, the sample was manually passed between a Gatan,\nIlion broad ion beam system and a dual beam FEI NOVA600i FIB-SEM. The milling\nconditions were 5 kV, 60 \uf06dA with 20 minutes mill time per slice, where the average slice\nthickness was 700 nm. After each material removal step a secondary electron (SE) image of\nthe surface was collected using the SEM at 0\u00b0 sample tilt, after which the crystallographic\ninformation is collected by EBSD mapping at 70\u00b0 sample tilt, to form SE-EBSD image pairs.\nThe quality of Ar+ ion polish achieved by with the Ilion BIB results in a very good surface for\nEBSD analysis without significant damage or phase transformation facilitating high indexing\nrates (above 90%). The co-registration of successive images was achieved using markers made\nby FIB [11]. SE-EBSD image pairs\n Presently, it is possible to undertake EBSD analysis for areas of several mm2\ndimensions by combining scanning of a large area by moving the electron beam and the\nstitching multiple scanned areas using the stage [6\u20138]. Although the size of the maps analysed\nhere are relatively modest compared to the largest ones reported elsewhere the distortion\ncorrections applied here are applicable to much larger areas too. The EBSD maps used in this\nwork were acquired at 200 nm step size with map size of 205 \u00d7 177 \uf06dm (1024 \u00d7 884 pixels)\nand were taken at 70\u00b0 angle of incidence and sample tilt using a Nordlys-S EBSD detector from\nOxford Instruments. This EBSD map size is relatively large for an ion based serial section\ntomography (SST) technique. The EBSD settings were chosen for the purpose of 3D-EBSD\nanalysis. Therefore, to ensure an accurate and high indexing percentage and to minimise mis-\nindexing for the WC and Co phases, a relatively high resolution EBSD detector setting of 4 \u00d7\n4 binning (336 \u00d7 256 pixels) was used, which means that the indexing time was about 25 ms\nper point, equivalent to 40 Hz, which was about 6 hours acquisition time per slice. Therefore\nthe EBSD settings used here resulted in an unusually slow speed for gathering an EBSD map\nper slice, however this meant that the average indexing rate was very high (> 96%) [11] and\nthe risk of low quality EBSD data was minimised for every slice. Furthermore, the slow EBSD\nmapping speed exaggerated any possible distortion effects from drift. Note that the Nordlys-S\nEBSD de\n\n## 5. stanek2020comparisonofsegmentation pages 1-2 (Context ID: pqac-00000010)\n\n Kovove Mater. 58 2020 301\u2013319 301\n DOI: 10.4149/km 2020 5 301\n\n\n\n Comparison of segmentation of 2D and 3D EBSD measurements\n in polycrystalline materials\n\n\n J. Stan\u02c7ek1*, J. Kope\u010dek2, P. Kr\u00e1l3, I. Kara\ufb01\u00e1tov\u00e14, F. Seitl4, V. Bene\u01614\n\n\n\n 1Department of Mathematics Education, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n 2Institute of Physics of the CAS, Department of Functional Materials, Na Slovance 2, 182 00 Prague, Czech Republic\n 3Institute of Physics of Materials, ASCR, \u02c7Zi\u017ekova 22, 616 62 Brno, Czech Republic\n 4Department of Probability and Mathematical Statistics, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n\n\n Received 30 December 2019, received in revised form 26 April 2020, accepted 17 June 2020\n\n\n Abstract\n\n A comparison study of grain segmentation in 2D and 3D electron backscatter di\ufb00rac-\n tion (EBSD) measurements of polycrystalline materials, based on a misorientation threshold,\n is presented. Using Laguerre tessellations with orientation marks, a simulation study that\n demonstrates a di\ufb00erent sensitivity of segmentation in the case of low angle and high angle\n grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investi-\n gated concerning segmentation, and there are given some recommendations for the misorien-\n tation thresholds. Computational tools such as DREAM 3D software enable signi\ufb01cant data\n evaluation that is completed by a \ufb01ve-parameter analysis of grain boundaries.\n\n\n K e y w o r d s: electron backscatter di\ufb00raction (EBSD), grain boundaries, misorientation,\n tessellation\n\n\n\n 1. Introduction observed the boundaries \u2013 the separators, whereas\n grains or rather their interiors were homogeneous ar-\n The electron backscattered di\ufb00raction method eas inside with de\ufb01ned or expected value of properties.\n(EBSD) has been a powerful tool in material science Of course, it is known that a grain structure with the\nsince the nineties, when a certain level of computeri- tessellation formula can be created [7]; nevertheless,\nzation and application of Hough transform took place, EBSD is the \ufb01nal step reversing axiom to construct\nsee [1, 2]. Nowadays, we can perform not only 2D the grains [2]. We have a set of measured points in the\nEBSD but exploiting focused ion beams (FIB), even EBSD method. We have positions of such points and\n3D EBSD, by materials\u2019 slicing tomography [3]. The their marks (e.g., Euler angles). Thus, here the inte-\n3D EBSD data evaluation is a complicated process [4]. rior is a quality, which exists, whereas the boundary\nStill, we expect to be able to construct grains [5] and has to be constructed. The easiest way how to create\nmicrostructure parameters by applying the same or grains from the set of measured points is the following\nsimilar procedures as in 2D EBSD data evaluation [6]. de\ufb01nition: the grain interior shall have the same or\nHowever, a discrepancy in the segmentation of data very similar properties. Thus, we construct grains by\npoints into the grains in the 2D and 3D cases is ap- collecting the measured points with the de\ufb01ned prop-\nparent, since a nonconvex grain in 3D can be observed erty. The metallographic and EBSD grains must be\nas more than one grain in a 2D slice. then the same in the optimal case. However, the re-\n The segmentation problem is deeply connected to sult of an experiment is not optimal. It contains some\nthe basic understanding of the polycrystals. Gener- noise or even a kind of systematic aberrations, and the\nally, the de\ufb01nition of grains as they were observed by reconstruction of grain structure can be done only up\nfounders of metallography was purely natural: They to some level of accuracy.\n\n*Corresponding author: tel.: +420 951553215; e-mail address: stanekj@karlin.m.cuni.cz302 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n The problem described above was omitted in com-\nputational materials science since the 3D data is fre-\nquently used in an elementary manner. Despite the\nexplosion of three-dimensional methods in the last\ndecades, the majority of 3D models use simple math-\nematical schemes to build the structure by a general\nprocedure approaching the reality by just a few pa-\nrameters, measured from 2D cuts of material [8].\n Data obtained by 3D EBSD or 2D EBSD form a\nset of orientations of measured points in a 3D ma-\ntrix (the points are called voxels in the sequel) or\n2D matrix (pixels in the sequel), respectively. The\nEBSD method considers a grain to be the set of all\npoints (voxels or pixels, respectively) with a similar\norientation, which are surrounded by a boundary of\na given misorientation threshold. The typical thresh-\nold in the EDAX OIM software is 5\u25e6[9], but other\nchoices are possible as well [10, 11]. The obtained seg-\nmentation into the grains depends on the chosen mis- Fig. 1. Di\ufb00erences between slices processed from 2D EBSD\n and the same slice processed from 3D EBSD. The greyorientation threshold, of course, but even if we use a\n colour displays the real grains (in the plane, which is or-\n\ufb01xed misorientation threshold, we may get di\ufb00erent\n thogonal to the cutting plane). The blue colour displays\nresults of the segmentation in 2D and 3D, as shown their 3D segmentation in the cutting plane, i.e., the parts\nin [12]. Therefore, the results obtained in 2D EBSD are considered to belong to the same grain. Finally, the\nmay not be easily compared with the results from 3D red and the orange colours show their 2D segmentation\nEBSD. in which they are considered to be di\ufb00erent grains. (a)\n In the paper, we present a comparison study of The situation with a nonconvex grain or a grain with a\nsegmentation of 2D and 3D EBSD measurements in boundary inside. (b) The situation with convex grains,\n boundaries with the misorientation smaller than the cho-polycrystalline materials, based on a misorientation\n sen threshold are denoted by the dashed line.\nthreshold. The main results of this topic are presented\nin Sections 2 and 3. The aim is to draw the reader\u2019s\nattention to the danger of obtaining misleading quan-\ntitative results about the geometry of microstructure crostructure analyses make use of DREAM 3D soft-\nwhen paying little attention to a proper segmentation. ware [14].\nFirst, in Section 2, the problem is demonstrated in a Some conclusions from Sections 2 and 3 are given\nsimulation study. A generalized Laguerre tessellation in more detail in the \ufb01nal part of the paper.\nmodeling of polycrystalline microstructure measured\nby 3D EBSD was used in [13]. We can simulate this\nmodel with a priori given distribution of misorienta- 2. Di\ufb00erences between 2D and 3D\ntions. The in\ufb02uence of this distribution on the seg- segmentation\nmentation is shown.\n In Section 3, we discuss the segmentation prob- The di\ufb00erences between segmentation into grains\nlem with two materials, namely the AlMgSc alloy af- in 2D and 3D are caused by the fact that in a 2D\nter ECAP and annealing, and pure copper after high- cutting plane, two sets of voxels may be surrounded\npressure torsion and annealing at room temperature. by a boundary de\ufb01ned by a given threshold of misori-\nThe polycrystalline microstructures are observed by entation so that the sets are detected as two grains,\n3D EBSD and 2D EBSD. The approaches presented while in 3D case, all these points belong to one larger\nin Section 2 for simulated toy data examples are dis- set which is surrounded by a boundary. Therefore in\ncussed subsequently for both materials. The curve of the 3D case, all these points lie in the same grain,\ndependence of the mean number of grains, in a single see Fig. 1. This behaviour can be observed when the\n2D slice, on the misorientation tolerance level, is plot- grains are nonconvex, or when there is a boundary of\nted for both 2D and 3D meth\n\n## 6. stanek2020comparisonofsegmentation pages 7-9 (Context ID: pqac-00000011)\n\negmentation\n with three misorientation thresholds is studied. The\n \ufb01rst one is 0.1\u25e6(called 0 % misorientation segmen-\n tation in the sequel), the second one is chosen so\n that 10 % of boundaries have lower misorientation\n (called 10 % misorientation segmentation in the se-\n quel). The last one corresponds to 15 % of boundaries\n with misorientation lower than the chosen threshold\n (called 15 % misorientation segmentation in the se-\n quel). The segmentation in 3D with such thresholds\n is shown in Fig. 8. The behaviour for all three cases\n (A), (C), and (D) is very similar. The 10 % misorien-\n tation segmentation contains larger grains, which are\n not presented in 0 % misorientation segmentation, and\n all three 15 % misorientation segmentations contain a\n single giant grain that covers most of the area of the\n slice. Note that the slice studied in the case (A) is\n the slice number 50, which has already been consid-\n ered above. Note that the results in 2D segmentation\n are similar for all three cases, i.e., the behaviour is\n similar to that one presented in Figs. 4a\u2013c. Therefore\n the segmentation in 2D is not visualized in this pa-\n per.\n Note that similar results are obtained when pro-\n cessing the other slices in the data set. Thus, based on\n this short simulation study, it can be concluded that\n the di\ufb00erence between processing of the data in 2D\n and 3D is signi\ufb01cant if at least 10\u201315 % of boundaries\n have misorientation lower than the threshold chosen\n for the segmentation procedure. The di\ufb00erence be-\n tween the processing of the data in 2D and 3D is sig-\nFig. 7. The simulated data: (a) case (C) and (b) case (D). ni\ufb01cantly a\ufb00ected by the thickness of the sample, but\n it seems that the morphology of the grains does not\n play an important part.\nated 25\u201374, and (f) segmentation in 3D with the entire\ndata set (100 slices). The result for the misorientation\nthreshold 2.5\u25e6is shown in Fig. 6. One can see that the 3. Real data\ndi\ufb00erence between the segmentation in 2D and 3D is\nmarginal when the specimen is thin (10 slices in our After the analysis of the simulated data in the pre-\ncase, i.e., 100 \u00d7 100 \u00d7 10 voxels), on the other hand, vious section, two real data sets are investigated, and\nthe di\ufb00erence is signi\ufb01cant when the specimen has 30 the segmentation into the grains in 2D and 3D is com-\nor more slices. The results are very similar if the data pared. The materials are described below in Sections\nset with 50 slices (e) or the entire data set with 100 3.1. and 3.2. It has to be added that most microstruc-\nslices (f) is used. tures show a particular orientation spread within the\n Another question is whether the grain size and grain interiors, and this is the result of the process-\nmorphology a\ufb00ect the di\ufb00erence between the segmen- ing history. It should be distinguished from acquisition\ntation in 2D and 3D as well. To investigate this de- noise, although for grain segmentation purposes, they\npendence, two more simulated data sets were added are usually not separated. Both materials have been\nto our study, namely (C) the Laguerre tessellation measured by the 3DEBSD method, which provides the\nwith a few distinctively larger grains and (D) the tes- data in the form of 2D slices. Such slices can be sub-\nsellation with nonconvex grains, see Fig. 7. The ori- sequently joined into a 3D image. Processing of the\nentation of the crystals in the cases (C) and (D) is data, such as removing noise, matching 3D slices into\nchosen in a similar way as in the case (A); there- a 3D image or removing grains, which are too small,\nfore, the misorientation between neighbouring grains etc., has been made by the software DREAM.3D. 2D\nis low, and the histograms of the misorientation are and 3D are compared in the same way as in the pre-\nsimilar to the histogram in Fig. 3a. The segmenta- vious section. First, a 2D slice processed separately308 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n\n\n\nFig. 8a\u2013f. The segmentation of the simulated data in 3D for the case (A) with the misorientation threshold (a) 0.1\u25e6, (b)\n2.39\u25e6, and (c) 2.58\u25e6. The segmentation of the simulated data in 3D for the case (C) with the misorientation threshold (d)\n 0.1\u25e6, (e) 2.22\u25e6, and (f) 2.34\u25e6.\n\n\n\nby DREAM.3D is chosen, and the obtained result is 3.1. Data of AlMgSc alloy\nquite similar to the results from working with 2D data.\nThen, the entire 3D data is processed in the same way The \ufb01rst investigated material is the aluminium\nand after 3D reconstruction, the chosen slice is cut alloy Al-3Mg-0.2Sc, which is an alloy known for its\nout. In this way, two processings of the same slice are superplasticity at high deformation rates [17]. The\nobtained (one in 2D and one in 3D), which are com- sample was subjected to severe plastic deformation\nparable. by equal channel angular pressing (ECAP), annealed J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 309\n\n\n\n\n\nFig. 8g\u2013i. The segmentation of the simulated data in 3D for the case (D) with the misorientation threshold (g) 0.1\u25e6, (h)\n 2.07\u25e6, and (i) 2.30\u25e6.\n\nsubsequently for 1 h at 400\u25e6C and investigated by\nthree-dimensional electron backscattered di\ufb00raction\n(3D-EBSD). Note that the size of the specimen was\n98.4 \u00d7 71.7 \u00d7 35.1 \u00b5m3 and it has been already investi-\ngated in detail in [12], where the di\ufb00erence between de-\ntection of grains in 2D and 3D is also brie\ufb02y discussed.\nIn the present paper, this di\ufb00erence is focused on more\ndetails. Since the sample includes a lot of noise, \ufb01lters\nin DREAM.3D have to be used to minimize the e\ufb00ect\nof the noise. The data as processed in DREAM.3D,\nwhere the grains are detected by using the \ufb01lter \u2018Seg-\nment Features (Misorientation)\u2019 with a misorientation\nthreshold of 2\u25e6, are visualized in Fig. 9. Note that the\nchoice of misorientation threshold as well as the set-\ntings for the procedure of removing noise \ufb01rmly in-\n\ufb02uence the number of detected grains. In our case, Fig. 9. The visualization of 3D data of AlMgSc alloy ex-\n cluding the noisy part of the specimen as well as very smallthe procedure detects 74 579 grains, but if we do not\n grains. The remaining part involves 74 579 grains.\nremove the noise and small grains (namely the grains\nconsisting of less than 8 voxels), the procedure detects\nmore than one million grains, where most of them are\nvery small (few voxels only). in a single slice is strongly a\ufb00ected by the noise, and\n Firstly, we compare the mean number of grains in a unfortunately, it is not clear how to remove this e\ufb00ect\nslice with the misorientation threshold chosen for the in the same way\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011]\n\nStatus: Paper Count=25 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=12 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_mH0dRIstCN2AOJxkl8ecCACI", + "info": { + "start_ts": 755.584416592, + "end_ts": 760.352480105 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_xTeD5YsxnoGXce4YcmR9STo0", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 1, \"updated_plan_objective_json\": \"{\\\"id\\\":1,\\\"objective\\\":\\\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\\\",\\\"rationale\\\":\\\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\\\",\\\"evaluation\\\":\\\"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\\\"}},{\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Error parsing plan step JSON: Failed to load JSON from text '{\"id\":1,\"objective\":\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\"rationale\":\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\"status\":\"completed\",\"result\":\"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\"evaluation\":\"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"}},{'.", + "name": "update_plan", + "tool_call_id": "call_xTeD5YsxnoGXce4YcmR9STo0", + "info": { + "start_ts": 783.853173669, + "end_ts": 783.855532371 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_Ky7P0IBBulkQ2t2Ljs7dEu3k", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 1, \"updated_plan_objective_json\": \"{\\\"id\\\":1,\\\"objective\\\":\\\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\\\",\\\"rationale\\\":\\\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\\\",\\\"evaluation\\\":\\\"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_Ky7P0IBBulkQ2t2Ljs7dEu3k", + "info": { + "start_ts": 805.401958358, + "end_ts": 805.403789393 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_ByYZSnq7nwSJDkOm9Cg48Zp8", + "type": "function", + "function": { + "arguments": "{\"artifact_type\": \"table\", \"description\": \"Create a concise prior-art evidence matrix with rows for the five requested areas. Columns: Area, closest published work(s), what is automated, objective/criterion, what is explicitly not automated, and gap assessment. Use only evidence already gathered. Include Polonsky 2019 grain segmentation parameter sweep; Shen 2019 CNN-confidence hybrid dictionary proposal; Zhu 2022 global optimization as a nearby per-pattern refinement work noting available metadata if evidence is limited; Sparks 2024 automated registration/pattern-center workflow; Burnett 2016 automated serial sectioning but manually/experimentally selected acquisition settings; Winiarski 2021 automated distortion correction but not indexing tuning; Kaufmann 2020 CNN phase mapping. Clearly distinguish demonstrated methods from proposals and from absent closed loops.\"}", + "name": "create_artifact" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "### Artifact (ID artifact-00)\n\n| Area | Closest published work(s) | What is automated | Objective/criterion | What is explicitly not automated | Gap assessment |\n|---|---|---|---|---|---|\n| 1. Optimizer-driven tuning of EBSD pattern-processing or indexing parameters | **Closest demonstrated EBSD parameter sweep:** Polonsky et al. 2019 parameterized DREAM.3D reconstruction variables and swept grain-segmentation misorientation tolerance over 0.1\u201310\u00b0 in 0.1\u00b0 increments, generating 100 reconstructions; **nearby but different:** Zhu et al. 2022 is only available here as metadata for \u201cglobal optimization algorithms\u201d applied to orientation/pattern-center/deformation extraction, so it should be treated as adjacent prior art, not verified from text in this record; **nearby non-tuning automation:** Kaufmann et al. 2020 CNN phase mapping automates phase recognition from patterns rather than tuning EBSD processing settings (pqac-00000014, pqac-00000013, pqac-00000004, pqac-00000005) | Automated batch execution / parameterization of reconstruction pipelines in BisQue+DREAM.3D; CNN-based direct phase classification from patterns in Kaufmann 2020 (pqac-00000014, pqac-00000004) | Polonsky used reconstruction-level metrics: fraction of \u201cgood\u201d CI-masked voxels retained and total grain count, selecting an elbow/balance point at 2.0\u00b0; Kaufmann targets phase-map fidelity qualitatively against known microstructure/EDS, not optimizer search (pqac-00000013, pqac-00000005) | No evidence here of Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, RL, random/grid search over **pattern** processing settings such as background correction, binning, neighbor averaging, Hough band count, rho-theta smoothing, or dictionary settings. Polonsky optimizes segmentation/reconstruction variables after indexing, not indexing settings themselves. Kaufmann does not tune EBSD parameters at all (pqac-00000013, pqac-00000005) | **Gap largely open.** Closest evidence is reconstruction-pipeline parameterization, not closed-loop optimizer-driven tuning of pattern-processing/indexing knobs. |\n| 2. Feedback loops that re-index stored raw patterns using CI/IQ/MAD/NCC to choose next parameter set | **Closest proposal:** Shen et al. 2019 proposes combining fast CNN predictions with a confidence measure (quaternion norm) so that high-confidence points get neighborhood dictionary search and low-confidence points get full dictionary search; **related raw-pattern availability, not loop:** Stinville et al. 2022 notes raw .up2 patterns can be opened in EMsoft DI / EMSphInx, enabling reindexing in principle (pqac-00000017, pqac-00000015, pqac-00000002, pqac-00000003) | Shen automates confidence estimation for each location and proposes routing subsequent dictionary matching effort based on confidence; raw pattern archival in Stinville enables offline reindexing workflows (pqac-00000017, pqac-00000003) | Quaternion norm as confidence proxy in Shen; no demonstrated CI/IQ/MAD/NCC-driven next-parameter selection loop in the gathered evidence (pqac-00000017) | Shen does **not** demonstrate an executed closed loop that iteratively changes indexing/processing parameters and re-indexes stored raw patterns to maximize CI/IQ/MAD/NCC/NCC-like objectives; it is a proposed hybrid search allocation strategy. Stinville provides data access, not autonomous optimization (pqac-00000017, pqac-00000003) | **Gap open.** Proposal-level hybrid confidence routing exists, but not a demonstrated feedback optimizer that selects the next EBSD parameter set from reindexed raw patterns. |\n| 3. Optimization against grain-level or microstructure-level outcomes | **Closest demonstrated:** Polonsky et al. 2019 explicitly optimizes a grain-segmentation parameter against microstructure-level outcomes (grain count and retained good-data fraction); Stan\u011bk et al. 2020 compares 2D/3D segmentation sensitivity to misorientation threshold and shows strong threshold dependence, but not autonomous optimization (pqac-00000013, pqac-00000012, pqac-00000010, pqac-00000011) | Automated parallel reconstructions and post-processing counts in Polonsky; analytical comparison of threshold effects on segmentation in Stan\u011bk (pqac-00000013, pqac-00000010) | Grain count stability / avoidance of grain merging plus retained CI-masked voxel fraction in Polonsky; sensitivity of number of grains and segmentation discrepancies vs threshold in Stan\u011bk (pqac-00000013, pqac-00000011) | No demonstrated multi-objective or constrained optimization over broader grain-level cleanup outcomes such as speckle/wild-grain fraction, grain-boundary integrity, fraction of pixels altered by cleanup, or Pareto balancing across multiple EBSD post-processing stages. Polonsky is a one-parameter sweep with manually interpreted elbow criterion (pqac-00000013, pqac-00000012) | **Partially occupied, but narrow.** Grain-level optimization exists for segmentation tolerance; broader autonomous multi-objective EBSD post-processing remains a gap. |\n| 4. Hybrid Hough-first / dictionary-on-residue and automated escalation criteria | **Closest published work:** Shen et al. 2019 explicitly proposes a hybrid strategy: CNN preliminary prediction plus confidence, then pattern matching only in a neighborhood for high-confidence points and full dictionary search for low-confidence points; this is the clearest automated escalation criterion in the gathered evidence. Sparks et al. 2024 and Winiarski et al. 2021 automate registration/distortion correction workflows but not Hough\u2192dictionary escalation (pqac-00000017, pqac-00000000, pqac-00000008) | Automated confidence scoring and proposed routing of dictionary-search scope in Shen; automated image registration/distortion correction in Sparks/Winiarski (pqac-00000017, pqac-00000000, pqac-00000008) | Criterion in Shen is confidence from quaternion norm: high confidence \u2192 local neighborhood dictionary search; low confidence \u2192 full dictionary search (pqac-00000017) | No evidence in the gathered text of a **demonstrated** Hough-first then dictionary-on-low-quality-residue loop with automated escalation based on CI/IQ/MAD/NCC thresholds. Shen\u2019s routing is proposed around CNN+dictionary, not Hough+dictionary, and not shown as an executed autonomous residue-reindexing pipeline (pqac-00000017) | **Mostly gap.** There is a close proposal for automated escalation, but not a demonstrated Hough-residue-to-dictionary autonomous workflow. |\n| 5. Autonomous/self-driving EBSD post-processing or characterization workflows, including links back to SEM acquisition recommendations | **Closest workflow automation:** Sparks et al. 2024 automates sample/image registration steps, pattern-center fitting for every section, and alignment routines; Winiarski et al. 2021 automates SE\u2013EBSD distortion correction and argues more automated procedures are needed; Burnett et al. 2016 demonstrates a fully automated serial-sectioning routine with tracking fiducials, but acquisition settings were experimentally chosen; Kaufmann et al. 2020 automates phase mapping from patterns (pqac-00000001, pqac-00000000, pqac-00000008, pqac-00000009, pqac-00000006, pqac-00000007, pqac-00000004) | Automated registration/warping, per-layer EBSD-to-BSE alignment, CI-threshold masks, pattern-center fitting per section, and automated serial sectioning with fiducial tracking; CNN inference for phase maps (pqac-00000000, pqac-00000001, pqac-00000006, pqac-00000004) | Workflow goals are geometric consistency, low inter-section misorientation, and corrected distortions; Burnett assesses equivalent quality via EBSD pattern quality/band contrast and indexing rate after experimentally selected settings (pqac-00000001, pqac-00000008, pqac-00000007) | No evidence here of a human-out-of-the-loop EBSD post-processing system that autonomously tunes post-processing/indexing parameters end-to-end, nor of a loop that feeds post-processing outcomes back to SEM acquisition parameters such as beam current, dwell time, or step size. Burnett explicitly determines acceptable milling current experimentally; Winiarski says more automated procedures still need to be developed (pqac-00000007, pqac-00000008) | **Gap open.** Sub-workflow automation exists, but not a self-driving closed-loop EBSD characterization system that recommends or controls acquisition parameters from post-processing outcomes. |\n\n\n*Table: This table maps the closest published prior art found for the five requested EBSD automation areas, distinguishing demonstrated methods from proposals and from absent closed loops. It highlights that most evidence clusters around workflow automation and grain-segmentation parameter sweeps, not autonomous closed-loop tuning of indexing/processing parameters.*\n\nArtifact IDs that may be injected into the answer: [artifact-00]\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013, pqac-00000014, pqac-00000015, pqac-00000016, pqac-00000017]\n\nStatus: Paper Count=25 | Relevant Papers=8 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=18 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "create_artifact", + "tool_call_id": "call_ByYZSnq7nwSJDkOm9Cg48Zp8", + "info": { + "start_ts": 829.394246022, + "end_ts": 852.626361711 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_sOxhMHmNJ6cGdwzj14gJbWOY", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 2, \"updated_plan_objective_json\": \"{\\\"id\\\":2,\\\"objective\\\":\\\"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\\\",\\\"rationale\\\":\\\"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\\\",\\\"evaluation\\\":\\\"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_Dgfm5HSnCqYdsd2pIw7qCf6a", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 3, \"updated_plan_objective_json\": \"{\\\"id\\\":3,\\\"objective\\\":\\\"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\\\",\\\"rationale\\\":\\\"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\\\",\\\"evaluation\\\":\\\"The evidence provides exact variables, criteria, and limitations.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_ERLgQRfcrgfEuXOF5b3WqhbB", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 4, \"updated_plan_objective_json\": \"{\\\"id\\\":4,\\\"objective\\\":\\\"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\\\",\\\"rationale\\\":\\\"Area 5 includes both software claims and closed-loop acquisition control.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\\\",\\\"evaluation\\\":\\\"Full text explicitly documents automation boundaries and calls for future automation.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_PqUIwWt0TdvA1AJFRXJAtK32", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 5, \"updated_plan_objective_json\": \"{\\\"id\\\":5,\\\"objective\\\":\\\"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\\\",\\\"rationale\\\":\\\"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\\\",\\\"evaluation\\\":\\\"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_Mv315OulvDSqQgHazJenPV4E", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 6, \"updated_plan_objective_json\": \"{\\\"id\\\":6,\\\"objective\\\":\\\"Create a structured artifact summarizing the evidence and limitations.\\\",\\\"rationale\\\":\\\"A compact evidence matrix will make the systematic survey auditable and useful.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Created artifact-00, a five-area evidence matrix with closest works, automation, objectives, exclusions, and gap assessments.\\\",\\\"evaluation\\\":\\\"Artifact covers all requested areas and clearly separates demonstrations, proposals, and absences.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_sOxhMHmNJ6cGdwzj14gJbWOY", + "info": { + "start_ts": 884.219772521, + "end_ts": 884.221180953 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_Dgfm5HSnCqYdsd2pIw7qCf6a", + "info": { + "start_ts": 884.221220607, + "end_ts": 884.221572424 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_ERLgQRfcrgfEuXOF5b3WqhbB", + "info": { + "start_ts": 884.221632058, + "end_ts": 884.221962904 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"completed\",\n \"result\": \"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\",\n \"evaluation\": \"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_PqUIwWt0TdvA1AJFRXJAtK32", + "info": { + "start_ts": 884.221985251, + "end_ts": 884.222258695 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"completed\",\n \"result\": \"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\",\n \"evaluation\": \"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a five-area evidence matrix with closest works, automation, objectives, exclusions, and gap assessments.\",\n \"evaluation\": \"Artifact covers all requested areas and clearly separates demonstrations, proposals, and absences.\"\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_Mv315OulvDSqQgHazJenPV4E", + "info": { + "start_ts": 884.22227648, + "end_ts": 884.222507565 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_JvmVAiIMF8mMq3af5Jp56p6p", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 7, \"updated_plan_objective_json\": \"{\\\"id\\\":7,\\\"objective\\\":\\\"Submit the final technical answer with evidence-matched citations.\\\",\\\"rationale\\\":\\\"Required final delivery step.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null}\"}", + "name": "update_plan" + } + } + ], + "info": null + } + ], + "hide_old_env_states": false, + "hide_old_action_content": false, + "sliding_window": null + }, + "next_agent_state": { + "tools": [ + { + "type": "function", + "info": { + "name": "paper_search", + "description": "Search for papers in Google Scholar for you to look at.\n\nCan be called repeatedly with identical parameters to retrieve more results (max 2 repeats).\nA short summary of the retrieved papers will be returned, but to examine the full content of each paper,\nyou must use the gather_evidence tool.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A Google Scholar search query, which can be a specific phrase, complete sentence,\nor general keywords, e.g. 'machine learning for immunology'. Also can be\ngiven search operators like 'author:'.", + "title": "Query", + "type": "string" + }, + "min_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for minimum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Min Year" + }, + "max_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for maximum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Max Year" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "gather_evidence", + "description": "Given a specific question, gather evidence from full-text of documents that have been retrieved by\nprevious tool calls.\n\nHere is how the tool works: the provided question is posed to all texts retrieved thus far\nby other tools. An LLM sees each text independently and is prompted to find information relevant\nto the question.\n\nThat is, we sample evidence `e \\sim LLM(prompt, text, q)` for each text. We only return pieces of\nevidence `e` that are sufficiently relevant to the question; if the text is irrelevant, then `e`\nis not included in the final response of this tool. Note that this LLM does not see the conversation\nhistory thus far, only the question and each text. Note that it only sees text, no images or other\nmedia from the contents (though tables and formulae may be parseable in some cases).\n\nThink of this tool as an efficient way to deeply scan all retrieved documents for evidence. It\nis complementary to the `read_text` tool, which can be used to directly inspect a subset of the\nretrieved documents.\n\nA valuable time to invoke this tool is after another tool finds potentially relevant documents.\nFeel free to invoke this tool in parallel with other tools, but do not call this tool in parallel with itself.\nOnly invoke this tool when the document count is above zero, or this tool will be useless.\n\nIf you call this tool 2+ times and repeatedly get 'No relevant evidence found',\nconsider whether the answer might be in visual content (figures/tables) rather than text,\nand use appropriate tools if available to access that content.\n\nOnce you receive evidence from this tool:\n\n1. Read Carefully: Don't skim. The evidence may contain multiple similar terms\n or values\u2014make sure you identify the RIGHT one.\n\n2. For Comparative Questions: If the question asks \"which X has greatest Y\",\n check that the evidence includes information about ALL relevant X options, not\n just one or two. If you only see partial information, gather more evidence.\n\n3. Check for Noise: Verify that numbers or terms aren't just metadata\n (page numbers, reference IDs, dates). Look for the value in meaningful\n scientific context.\n\n4. Context Matters: A term appearing 71 times doesn't mean it's always the answer.\n Verify which occurrence actually answers the question asked.\n\nIf the evidence doesn't clearly answer the question, gather more evidence before\nconcluding. Don't guess based on partial information.", + "parameters": { + "type": "object", + "properties": { + "question": { + "description": "Specific question to gather evidence for.", + "title": "Question", + "type": "string" + }, + "text_name_focus_list": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "if there are known text names that are relevant to the question,\nyou can specify them here. This will ensure that the tool will look at these text names\nfirst, and then look at the rest of the evidence. If left as `None`, an embedding\nsearch will be performed on underlying content to find most relevant texts.", + "title": "Text Name Focus List" + } + }, + "required": [ + "question" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "open_targets_search", + "description": "Search for disease-target associations from OpenTargets database.\n\nAdds new disease-target association data to the state and returns top search results.", + "parameters": { + "type": "object", + "properties": { + "disease_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of disease names to search for.\nUsage:\ndisease_names = [\"Alzheimer's\"] ->\n Returns top targets for Alzheimer's\ndisease_names = [\"Alzheimer's\", \"Diabetes\"], target_names = [\"PSEN1\"] ->\n Filters for only PSEN1-Alzheimer's/Diabetes evidence\ndisease_names = [\"Alzheimer's\"], drug_names = [\"Aducanumab\"] ->\n Filters for Alzheimer's disease-target pairs treated by Aducanumab", + "title": "Disease Names" + }, + "target_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of target names to search for.\nUsage:\ntarget_names = [\"BACE1\"] ->\n Returns top diseases associated with BACE1\ntarget_names = [\"BACE1\", \"APP\"], disease_names = [\"Alzheimer's\"] ->\n Filters for only these target-Alzheimer's pairs\ntarget_names = [\"BACE1\"], drug_names = [\"Aducanumab\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Target Names" + }, + "drug_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of drug names to generate or filter results.\nUsage:\ndrug_names = [\"Aducanumab\"] ->\n Returns all disease-target pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], disease_names = [\"Alzheimer's\"] ->\n Filters for Alzheimer's pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], target_names = [\"BACE1\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Drug Names" + } + }, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "collect_cited_papers_in_evidence", + "description": "Collect papers by traversing the citations of relevant papers to increase the paper count.\n\nThis tool has no effect if called when paper count or relevant papers are zero.\nThis tool will find papers that are likely to lead to more relevant evidence.\nThe papers found have a higher chance of relevance because this tool starts from relevant evidence.\nThis tool's usefulness increases with the amount of gathered evidence as there's more citations to consider.", + "parameters": { + "type": "object", + "properties": {}, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "view_images", + "description": "Retrieve images from a document.\n\nUse this tool when you need to see figures, tables, or other visual content.\nThe tool works by asking a weaker LLM agent to scan through the document, looking for\nvisual content matching the query you provide. When the agent finds matches, it will\ncrop the page to the relevant region. It can be tasked to return multiple images,\nif desired. Note that the weaker LLM may be worse at *understanding* images than\nyou, so double-check its conclusions about the returned images. It is also stateless,\ni.e. it will not remember what you have asked it to find in the past.\n\nUse this tool carefully. It is difficult to retrieve images, and the images\nthemselves will fill up your context window quickly. Here are some guidelines:\n1. ALWAYS use this tool if the question explicitly mentions 'Table X' or 'Figure Y'.\n These questions require visual content and cannot be answered with text alone.\n2. Use this tool as a fallback when gather_evidence fails repeatedly (2+ times\n returning 'No relevant evidence'). This suggests the answer may be in visual content.\n3. When gather_evidence returns text mentioning 'see Figure X' or 'shown in Table Y',\n you should call this tool to examine that specific visual reference.\n4. Only call this tool after directly inspecting relevant text with read_text\n to confirm the document likely contains the visual you need.\n5. Don't use this tool in a broad-spectrum way (don't call on many documents\n just to see what might be relevant).\n6. Never call this tool in parallel with itself.\n7. Users like visual evidence, so call this tool at least once before answering so you\n can provide some image citations in the answer. Add this to your plan so you do not forget.\n For images to be presented to the user, they must be cited in the final answer after retrieval\n by this tool. You may ignore this instruction if the request is for the question is asking\n for a specific piece of information and is not an open-ended research question.\n8. NEVER cite full-page images in your final answer, unless they are full-page panels figures or\n tables. Cite relevant cropped regions if available, otherwise consider calling this tool again\n to find a smaller region that answers the query.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "Query to search for visual content in the document. If you know the specific\nfigure/table/formula you are looking for, you can specify it here (e.g. \"find me\nFigure 3\" or \"extract the table on page 10\"). You can also request based on content\n(e.g. \"find me all volcano plots of gene X\" or \"I need the formula for the growth rate\").\nYou can even include explicit instructions for the LLM agent to follow (if helpful, not\nnecessarily required). The more specific and clear your query is here, the higher the\nchance of success.", + "title": "Query", + "type": "string" + }, + "text_name": { + "description": "Name of a text chunk (e.g., 'smith2023example pages 1-5').", + "title": "Text Name", + "type": "string" + } + }, + "required": [ + "query", + "text_name" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "create_artifact", + "description": "Create an artifact that can be used in the final answer, based on information\nyou have collected so far.\n\nHow it works: the full message history thus far is passed to another LLM,\nalong with the question and your description. The LLM is tasked to create\nan artifact that matches your specifications, including a title, caption,\nand markdown content. You drive the process by specifying the type and\ndescribing what you want.\n\nOnly use this tool *after* you have gathered sufficient evidence or\ninformation to create a new artifact. Do not call this tool in parallel\nwith tools that are searching for information or evidence that would be\nnecessary to create this artifact.\n\nIt is very helpful after calls to several search tools,\nafter calls to the `gather_evidence` tool, but it is not required.\nHowever, it should be called before the `answer` tool at\nleast one time, but more may be given if specifically helpful for\nthe task, so that the `answer` tool can use the artifacts for the answer.\n\nAn artifact can be any string representable fact or list of facts,\nformatted using markdown syntax. The artifact will be generated as one of\nthe following types:\n- table: Markdown tables with rows and columns (e.g., findings, methods, molecules)\n- code_block: Code snippets in fenced code blocks\n- blockquote: Quoted text or important facts (using > prefix)\n- image: AI-generated image (chart, diagram, or visualization)\n\nEach artifact consists of:\n- A short title (3-8 words) that names what the artifact contains\n- A caption (1-2 sentences) describing what the artifact shows and why it's useful\n- Markdown content in a single element type (validated automatically with retries if needed)\n For image type, content is a markdown image embed ``![title](data:image/png;base64,...)``,\n embedding the PNG directly as a base64 data URL.", + "parameters": { + "type": "object", + "properties": { + "artifact_type": { + "description": "Artifact type to create. Must be one of: table,\ncode_block, blockquote, image.", + "title": "Artifact Type", + "type": "string" + }, + "description": { + "description": "Description of what you want in the artifact. Be specific\nabout what information should be included, how it should be organized,\nand what columns/sections/content you want. This guides the summary LLM.\nFor image type, describe the visual you want generated in detail.", + "title": "Description", + "type": "string" + } + }, + "required": [ + "artifact_type", + "description" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "update_plan", + "description": "Update the plan for a specific or new objective using a JSON string to create or\nreplace the existing objective.\n\nThe updated_json should be a JSON string representing a single plan objective object.\n\nYou may use this tool to update an objective, create a new one, or change\nstrategy if the current plan is not working. Only change strategy significantly\nas a last resort, after trying very hard to make the current plan work.\n\nOnly mark an objective as completed if you are absolutely sure that it has been fully\nmet. If it is only partially met, you may add additional objectives to the plan for\nwhat is missing.", + "parameters": { + "type": "object", + "properties": { + "plan_objective_id": { + "description": "The id of the objective to update.", + "title": "Plan Objective Id", + "type": "integer" + }, + "updated_plan_objective_json": { + "description": "A JSON string representing the updated objective.\nIt should be a dictionary with the with the following keys:\n- \"id\": (int) an integer id matching plan_objective_id, if not found, a new objective is created.\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Updated Plan Objective Json", + "type": "string" + } + }, + "required": [ + "plan_objective_id", + "updated_plan_objective_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "clinical_trials_search", + "description": "Search for clinical trials, with support for repeated calls and concurrent execution.\n\nWill add new clinical trials to the state, and return metadata about the number of trials found.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "The search query string. Supports complex boolean expressions, field-specific\nsearches, and query modifiers through operators. All configuration is done through\noperators in the query string.\nQuery Syntax:\n Basic Search:\n Simple text automatically uses default EXPANSION[Relaxation] and COVERAGE[Contains]\n >>> \"heart attack\"\n\n Modified Search:\n Use operators to modify search behavior:\n >>> 'EXPANSION[None]COVERAGE[FullMatch]\"exact phrase\"'\n >>> 'EXPANSION[Concept]heart attack'\n\n Field Search:\n Specify fields using AREA operator:\n >>> 'AREA[InterventionName]aspirin'\n >>> 'AREA[Phase]PHASE3'\n\n Location Search:\n Use SEARCH operator for compound location queries:\n >>> 'cancer AND SEARCH[Location](AREA[LocationCity]Boston AND AREA[LocationState]Massachusetts)'\n\n Complex Boolean:\n Combine terms with AND, OR, NOT and parentheses:\n >>> '(cancer OR tumor) AND NOT (EXPANSION[None]pediatric OR AREA[StdAge]CHILD)'\n\n Date Ranges:\n Use RANGE to specify date ranges with formats like \"yyyy-MM\" or \"yyyy-MM-dd\".\n Note that MIN and MAX can be used for open-ended ranges:\n >>> AREA[ResultsFirstPostDate]RANGE[2015-01-01, MAX]\n\nOperators:\n EXPANSION[type]: Controls term expansion\n - None: Exact match only, case and accent sensitive\n - Term: Includes lexical variants (plurals, spellings)\n - Concept: Includes UMLS synonyms\n - Relaxation: Relaxes adjacency requirements (default)\n - Lossy: Allows missing partial terms\n\n COVERAGE[type]: Controls text matching\n - FullMatch: Must match entire field\n - StartsWith: Must match beginning of field\n - EndsWith: Must match end of field\n - Contains: Must match part of field (default)\n\n AREA[field]: Specifies field to search\n - See Field Reference for available fields\n\n SEARCH[type]: Groups field searches\n - Location: Groups location-related fields\n - Study: Groups study-related fields\n\nUsage Notes:\n - All search expressions are implicitly OR expressions\n - Operator precedence (highest to lowest): terms/source operators, NOT/context operators, AND, OR\n - Use quotes for exact phrase matching: \"heart attack\"\n - Use parentheses for grouping: (heart OR cardiac) AND attack\n - Use backslash to escape operators: \\AND\n - Default expansion is EXPANSION[Relaxation]\n - Default coverage is COVERAGE[Contains]\n\nField Reference:\n High Priority Fields (weight >= 0.8):\n - NCTId (1.0): Trial identifier\n - Acronym (1.0): Study acronym\n - BriefTitle (0.89): Short title\n - OfficialTitle (0.85): Full official title\n - Condition (0.81): Medical condition\n - InterventionName (0.8): Primary intervention name\n - OverallStatus: Trial status\n\n Medium Priority Fields (0.5-0.79):\n - InterventionOtherName (0.75): Alternative intervention names\n - Phase (0.65): Trial phase\n - StdAge (0.65): Standard age groups\n - Keyword (0.6): Study keywords\n - BriefSummary (0.6): Short description\n - SecondaryOutcomeMeasure (0.5): Secondary outcomes\n\n Low Priority Fields (< 0.5):\n - DesignPrimaryPurpose (0.3): Primary purpose of study\n - StudyType (0.3)\n - Various descriptive, location, and administrative fields\n\nSupported Enums:\n Phase:\n - EARLY_PHASE1: Early Phase 1\n - PHASE1: Phase 1\n - PHASE2: Phase 2\n - PHASE3: Phase 3\n - PHASE4: Phase 4\n - NA: Not Applicable\n\n StandardAge:\n - CHILD: Child\n - ADULT: Adult\n - OLDER_ADULT: Older Adult\n\n Status:\n - RECRUITING: Currently recruiting participants\n - ACTIVE_NOT_RECRUITING: Active but not recruiting\n - COMPLETED: Study completed\n - ENROLLING_BY_INVITATION: Enrolling by invitation only\n - NOT_YET_RECRUITING: Not yet recruiting\n - SUSPENDED: Study suspended\n - TERMINATED: Study terminated\n - WITHDRAWN: Study withdrawn\n - AVAILABLE: Available\n - NO_LONGER_AVAILABLE: No longer available\n - TEMPORARILY_NOT_AVAILABLE: Temporarily not available\n - APPROVED_FOR_MARKETING: Approved for marketing\n - WITHHELD: Withheld\n - UNKNOWN: Unknown status\n\n StudyType:\n - INTERVENTIONAL: Interventional studies\n - OBSERVATIONAL: Observational studies\n - EXPANDED_ACCESS: Expanded access studies\n\n PrimaryPurpose:\n - TREATMENT: Treatment\n - PREVENTION: Prevention\n - DIAGNOSTIC: Diagnostic\n - ECT: Educational/Counseling/Training\n - SUPPORTIVE_CARE: Supportive Care\n - SCREENING: Screening\n - HEALTH_SERVICES_RESEARCH: Health Services Research\n - BASIC_SCIENCE: Basic Science\n - DEVICE_FEASIBILITY: Device Feasibility\n - OTHER: Other\n\n InterventionType:\n - BEHAVIORAL: Behavioral interventions\n - BIOLOGICAL: Biological interventions\n - COMBINATION_PRODUCT: Combination product interventions\n - DEVICE: Device interventions\n - DIAGNOSTIC_TEST: Diagnostic test interventions\n - DIETARY_SUPPLEMENT: Dietary supplement interventions\n - DRUG: Drug interventions\n - GENETIC: Genetic interventions\n - PROCEDURE: Procedure interventions\n - RADIATION: Radiation interventions\n - OTHER: Other interventions\n\n DesignAllocation:\n - RANDOMIZED: Randomized allocation\n - NON_RANDOMIZED: Non-randomized allocation\n - NA: Not applicable\n\n InterventionalAssignment:\n - SINGLE_GROUP: Single group assignment\n - PARALLEL: Parallel assignment\n - CROSSOVER: Crossover assignment\n - FACTORIAL: Factorial assignment\n - SEQUENTIAL: Sequential assignment\n\n ObservationalModel:\n - COHORT: Cohort\n - CASE_CONTROL: Case-Control\n - CASE_ONLY: Case-Only\n - CASE_CROSSOVER: Case-Crossover\n - ECOLOGIC_OR_COMMUNITY: Ecologic or Community\n - FAMILY_BASED: Family-Based\n - DEFINED_POPULATION: Defined Population\n - NATURAL_HISTORY: Natural History\n - OTHER: Other\n\n DesignMasking:\n - NONE: None (Open Label)\n - SINGLE: Single\n - DOUBLE: Double\n - TRIPLE: Triple\n - QUADRUPLE: Quadruple\n\n WhoMasked:\n - PARTICIPANT: Participant\n - CARE_PROVIDER: Care Provider\n - INVESTIGATOR: Investigator\n - OUTCOMES_ASSESSOR: Outcomes Assessor", + "title": "Query", + "type": "string" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "create_plan", + "description": "Create a plan for to construct a thorough and accurate answer.\n\nDecompose the input task into a set of sub-objectives that need to\nbe met in order to form the answer. Subsequently, when working on solving\nthe query, you can focus on these objectives one by one. Note that the\n`update_plan` tool allows you to change the objectives if needed.\n\nIn addition to the objectives you deem necessary, add one final objective\nto call the `answer` tool with your final answer, so that you do not\nforget to submit your answer.", + "parameters": { + "type": "object", + "properties": { + "plan_json": { + "description": "A JSON string representing the plan. It should be a dictionary\nwith top-level key \"objectives\", whose value is a list of objectives.\nEach objective should be a dictionary with the following keys:\n- \"id\": (int) an integer id for the plan objective\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Plan Json", + "type": "string" + } + }, + "required": [ + "plan_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "read_text", + "description": "Given a list of text names, read their contents.\n\nThis tool is valuable to read the full contents of specific parts of documents,\ncomplementing the `gather_evidence` tool's indirect approach.\nThis can be an especially useful tool if `gather_evidence` does not find evidence,\nbut you think it is very likely that specific documents are still relevant to the question.", + "parameters": { + "type": "object", + "properties": { + "text_name_list": { + "description": "A list of up to 10 text names to read.\nThese text names can be any valid text names retrieved by other tools.", + "items": { + "type": "string" + }, + "title": "Text Name List", + "type": "array" + } + }, + "required": [ + "text_name_list" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "patent_search", + "description": "Search for patents to increase the patent count.\n\nRepeat previous calls with the same query to continue a search.\nOnly repeat a maximum of three times.\nThis tool can be called concurrently.\nThis tool introduces novel patents, so invoke it when just beginning\nor when unsatisfied with the current evidence.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A search query for patents (using Google Patents). Never include temporal\nfilters here (e.g. specific years or dates); instead use the `after` and `before`\narguments. Never include author filters here; instead use the `inventor` argument.\nSupports the following syntax, but they can be brittle and are not preferred; try\nsimple queries where possible:\n- Boolean operators: AND (default), OR, and - (NOT)\n- Exact phrases: \"quoted text\"\n- Proximity: NEAR/x, ADJ/x (ordered), WITH (20 words), SAME (200 words)\n- Field search: TI=(title), AB=(abstract), CL=(claims)\n- CPC codes: cpc:A01B or CPC=B60R22/low (includes children)\n- Wildcards: ? (0-1 char), * (0+ chars), # (exactly 1 char)\n- Chemistry: exact molecule names, SSS=molecule (substructure),\n ~molecule (similar), SMILES strings, SMARTS=pattern, InChI keys,\n CL=~molecule (chemistry in claims)", + "title": "Query", + "type": "string" + }, + "after": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Minimum patent date. Format: YYYYMMDD.\nExample: '20150101' for patents after Jan 1, 2015.", + "title": "After" + }, + "before": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Maximum patent date. Format: YYYYMMDD.\nExample: '20200101' for patents before Jan 1, 2020.", + "title": "Before" + }, + "date_type": { + "default": "publication", + "description": "Type of date that after and before refer to. Must be either 'publication'\nor 'filing'. Only necessary to set if after and/or before are set.", + "title": "Date Type", + "type": "string" + }, + "inventor": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Filter by inventor name(s). Comma-separated for multiple.", + "title": "Inventor" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "answer", + "description": "Submit your final answer to the user. You MUST call this tool on your final answer.\nIf you mark all objectives as completed but have not yet called this tool, then\nthe answer will not be submitted. Note that this tool should only be called once, as it\nimmediately terminates the run and submits the answer.\n\nWrite an answer based on the evidence from gather_evidence results in message history.\nIf the context provides insufficient information, provide a partial answer and explain\nwhy it is partial (e.g. the question is currently an open question in the field,\nyou were unable to obtain certain papers, etc.).\n\nFor each part of your answer, indicate which sources most support it via citation IDs at\nthe end of sentences, like (pqac-0000ABCD).\nYou are only allowed to cite available context IDs, indicated in previous messages by \"Context IDs that may be cited:\".\n## Valid citation examples, only use comma/space delimited parentheticals:\n\n- (pqac-0000ABCD, pqac-0000EFGH)\n- (pqac-0000IJKL)\n\n## Invalid citation examples:\n\n- (pqac-0000ABCD and pqac-0000EFGH)\n- (pqac-0000ABCD;pqac-0000EFGH)\n- (pqac-0000ABCD-pqac-0000EFGH)\n- pqac-0000ABCD and pqac-0000EFGH\n- Example's work pqac-0000ABCD\n- (pages pqac-0000ABCD)\n- ![description](pqac-0000ABCD)\n- ()\n\nDo not concatenate citation ids, just use them as-is.\n\nWhen writing your answer, embed artifacts if relevant artifacts have been previously\ncreated by the `create_artifact` tool. To embed an artifact, simply include this in your\nanswer, on its own line and without any additional formatting (i.e. no backticks, code fences,\nquotes, additional tags, etc.).\nAlways provide at least some introductory sentence or paragraph before referring to an artifact,\nso that the artifact is properly contextualized.\nHere is an example reference to an artifact with ID \"artifact-XY\":\n\n\nValid artifact IDs are indicated in previous messages by 'Artifact IDs that may be injected into the answer:'.\n\n\nDouble-check your answer:\n\n1. Evidence Match: Re-read the specific evidence snippets you're citing. Does your\n answer actually match what they say? Don't rely on memory from 20+ messages ago.\n\n2. For Comparative Questions (\"which X has greatest/maximum/highest Y\"):\n - Did you examine ALL options mentioned in the evidence?\n - Did you explicitly compare their values?\n - Example: If evidence mentions \"Pancreatic: 0.92, Liver: 0.85, Lung: 0.78\",\n verify you compared all three, not just two.\n\n3. For Specific Value Questions (\"How many\", \"What is the\", specific terms):\n - Does the exact value you're stating appear in the evidence?\n - Are you extracting from the right context (not noise like page numbers)?\n\n4. Sanity Check: Read your answer. Does it directly address what was asked?\n\nDon't use Markdown image syntax like ![...](...) to reference media in your answer,\nbecause media get automatically linked by the system based on your answer's citations.\n\nWrite in the style of a scientific article, with concise sentences and\ncoherent paragraphs. This answer will be used directly,\nso do not add any extraneous information.", + "parameters": { + "type": "object", + "properties": { + "answer": { + "description": "valid answer, cited with citation IDs.", + "title": "Answer", + "type": "string" + } + }, + "required": [ + "answer" + ] + } + } + } + ], + "messages": [ + { + "role": "user", + "content": "Use the tools to answer the question: Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.\n\nStart by making a plan, and update the plan as you go. Your answer will be returned to the user when all plan objectives are completed. Be thorough and ensure your evidence matches the user's question.", + "info": null + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_TMwBuCDwZNGWGVmRuRrINX4G", + "type": "function", + "function": { + "arguments": "{\"plan_json\": \"{\\\"objectives\\\":[{\\\"id\\\":1,\\\"objective\\\":\\\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\\\",\\\"rationale\\\":\\\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":2,\\\"objective\\\":\\\"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\\\",\\\"rationale\\\":\\\"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":3,\\\"objective\\\":\\\"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\\\",\\\"rationale\\\":\\\"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":4,\\\"objective\\\":\\\"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\\\",\\\"rationale\\\":\\\"Area 5 includes both software claims and closed-loop acquisition control.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":5,\\\"objective\\\":\\\"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\\\",\\\"rationale\\\":\\\"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":6,\\\"objective\\\":\\\"Create a structured artifact summarizing the evidence and limitations.\\\",\\\"rationale\\\":\\\"A compact evidence matrix will make the systematic survey auditable and useful.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":7,\\\"objective\\\":\\\"Submit the final technical answer with evidence-matched citations.\\\",\\\"rationale\\\":\\\"Required final delivery step.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null}]}\"}", + "name": "create_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Plan registered and formatted correctly, with some modifications:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "create_plan", + "tool_call_id": "call_TMwBuCDwZNGWGVmRuRrINX4G", + "info": { + "start_ts": 386.692433131, + "end_ts": 386.695581364 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_9At2TUTAjnJ23z8NGaLTDASo", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated optimization indexing parameters Hough Bayesian optimization genetic algorithm grid search\"}", + "name": "paper_search" + } + }, + { + "id": "call_5aBBEXJupU0KmlHg00b79lgu", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD reindex raw patterns confidence index image quality objective optimization parameter tuning\"}", + "name": "paper_search" + } + }, + { + "id": "call_acsPy2saVbrPRcVvQrHXhwqx", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD hybrid Hough dictionary indexing low confidence patterns automated\"}", + "name": "paper_search" + } + }, + { + "id": "call_SvSRR7s0yK3PpnhQQ0Vw5Xl4", + "type": "function", + "function": { + "arguments": "{\"query\": \"autonomous self-driving EBSD acquisition post-processing SEM workflow\"}", + "name": "paper_search" + } + }, + { + "id": "call_yBdeyyzRACDmvHLG4V5YZoxC", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD grain reconstruction parameter optimization misorientation threshold wild grains cleanup automated\"}", + "name": "paper_search" + } + }, + { + "id": "call_4uZ0w5ewQsn4Ra9JRMzH48rh", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated parameter optimization indexing Hough dictionary feedback\"}", + "name": "patent_search" + } + }, + { + "id": "call_WJzVeGxOMudv7uXJmeYdSe3F", + "type": "function", + "function": { + "arguments": "{\"query\": \"electron backscatter diffraction autonomous acquisition processing\"}", + "name": "patent_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Emerging materials intelligence ecosystems propelled by machine learning\nValid Text Names:\n'batra2021emergingmaterialsintelligence pages 1-2', 'batra2021emergingmaterialsintelligence pages 2-3', 'batra2021emergingmaterialsintelligence pages 3-4', 'batra2021emergingmaterialsintelligence pages 4-5', 'batra2021emergingmaterialsintelligence pages 5-6', 'batra2021emergingmaterialsintelligence pages 6-7', 'batra2021emergingmaterialsintelligence pages 7-8', 'batra2021emergingmaterialsintelligence pages 8-9', 'batra2021emergingmaterialsintelligence pages 9-10', 'batra2021emergingmaterialsintelligence pages 10-11', 'batra2021emergingmaterialsintelligence pages 11-12', 'batra2021emergingmaterialsintelligence pages 12-13', 'batra2021emergingmaterialsintelligence pages 13-14', 'batra2021emergingmaterialsintelligence pages 14-15', 'batra2021emergingmaterialsintelligence pages 15-16', 'batra2021emergingmaterialsintelligence pages 16-17', 'batra2021emergingmaterialsintelligence pages 17-18', 'batra2021emergingmaterialsintelligence pages 18-19', 'batra2021emergingmaterialsintelligence pages 19-20', 'batra2021emergingmaterialsintelligence pages 20-21', 'batra2021emergingmaterialsintelligence pages 21-22', 'batra2021emergingmaterialsintelligence pages 21-21', 'batra2021emergingmaterialsintelligence pages 22-22', 'batra2021emergingmaterialsintelligence pages 23-23', 'batra2021emergingmaterialsintelligence pages 23-24'\n\nBibTex:\n@article{batra2021emergingmaterialsintelligence,\n author = \"Batra, Rohit and Song, Le and Ramprasad, Rampi\",\n title = \"Emerging materials intelligence ecosystems propelled by machine learning\",\n year = \"2021\",\n journal = \"Nature Reviews Materials\",\n volume = \"6\",\n pages = \"655-678\",\n month = \"Nov\",\n doi = \"10.1038/s41578-020-00255-y\",\n url = \"https://doi.org/10.1038/s41578-020-00255-y\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"8\",\n issn = \"2058-8437\"\n}\n\n\nRelevant Snippet:\n\u2026 For the first point, Bayesian optimization and genetic algorithms have been successfully \u2026 than traditional indexing methods that are either susceptible to noise (Hough transformation \u2026\n\n\n---\n\n## 2. Materials Informatics for Mechanical Deformation: A Review of Applications and Challenges\nValid Text Names:\n'frydrych2021materialsinformaticsfor pages 1-2', 'frydrych2021materialsinformaticsfor pages 2-4', 'frydrych2021materialsinformaticsfor pages 4-5', 'frydrych2021materialsinformaticsfor pages 5-7', 'frydrych2021materialsinformaticsfor pages 7-8', 'frydrych2021materialsinformaticsfor pages 8-10', 'frydrych2021materialsinformaticsfor pages 10-12', 'frydrych2021materialsinformaticsfor pages 12-13', 'frydrych2021materialsinformaticsfor pages 13-15', 'frydrych2021materialsinformaticsfor pages 15-17', 'frydrych2021materialsinformaticsfor pages 17-19', 'frydrych2021materialsinformaticsfor pages 19-20', 'frydrych2021materialsinformaticsfor pages 20-22', 'frydrych2021materialsinformaticsfor pages 22-23', 'frydrych2021materialsinformaticsfor pages 23-24', 'frydrych2021materialsinformaticsfor pages 24-26', 'frydrych2021materialsinformaticsfor pages 26-27', 'frydrych2021materialsinformaticsfor pages 27-28', 'frydrych2021materialsinformaticsfor pages 28-30', 'frydrych2021materialsinformaticsfor pages 30-31'\n\nBibTex:\n@article{frydrych2021materialsinformaticsfor,\n author = \"Frydrych, Karol and Karimi, Kamran and Pecelerowicz, Michal and Alvarez, Rene and Dominguez-Guti\u00e9rrez, Francesco Javier and Rovaris, Fabrizio and Papanikolaou, Stefanos\",\n title = \"Materials Informatics for Mechanical Deformation: A Review of Applications and Challenges\",\n year = \"2021\",\n journal = \"Materials\",\n volume = \"14\",\n pages = \"5764\",\n month = \"Oct\",\n doi = \"10.3390/ma14195764\",\n url = \"https://doi.org/10.3390/ma14195764\",\n publisher = \"MDPI AG\",\n issue = \"19\",\n issn = \"1996-1944\"\n}\n\n\nAbstract:\nIn the design and development of novel materials that have excellent mechanical properties, classification and regression methods have been diversely used across mechanical deformation simulations or experiments. The use of materials informatics methods on large data that originate in experiments or/and multiscale modeling simulations may accelerate materials\u2019 discovery or develop new understanding of materials\u2019 behavior. In this fast-growing field, we focus on reviewing advances at the intersection of data science with mechanical deformation simulations and experiments, with a particular focus on studies of metals and alloys. We discuss examples of applications, as well as identify challenges and prospects.\n\nRelevant Snippet:\n\u2026 the speed of indexing and classification of EBSD patterns as \u2026 EBSD data. For example, ML methods can be used [75] to \u2026 [78] based on an evolutionary algorithm (EA) may be applied, \u2026\n\n\n---\n\n## 3. Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields\nValid Text Names:\n'stinville2022multimodaldatasetof pages 1-2', 'stinville2022multimodaldatasetof pages 2-3', 'stinville2022multimodaldatasetof pages 3-5', 'stinville2022multimodaldatasetof pages 5-7', 'stinville2022multimodaldatasetof pages 7-9', 'stinville2022multimodaldatasetof pages 9-12', 'stinville2022multimodaldatasetof pages 12-13', 'stinville2022multimodaldatasetof pages 13-15', 'stinville2022multimodaldatasetof pages 15-15', 'stinville2022multimodaldatasetof pages 15-16'\n\nBibTex:\n@article{stinville2022multimodaldatasetof,\n author = \"Stinville, J. C. and Hestroffer, J. M. and Charpagne, M. A. and Polonsky, A. T. and Echlin, M. P. and Torbet, C. J. and Valle, V. and Nygren, K. E. and Miller, M. P. and Klaas, O. and Loghin, A. and Beyerlein, I. J. and Pollock, T. M.\",\n title = \"Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields\",\n year = \"2022\",\n journal = \"Scientific Data\",\n volume = \"9\",\n month = \"Aug\",\n doi = \"10.1038/s41597-022-01525-w\",\n url = \"https://doi.org/10.1038/s41597-022-01525-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2052-4463\"\n}\n\n\nAbstract:\nAbstractThe development of high-fidelity mechanical property prediction models for the design of polycrystalline materials relies on large volumes of microstructural feature data. Concurrently, at these same scales, the deformation fields that develop during mechanical loading can be highly heterogeneous. Spatially correlated measurements of 3D microstructure and the ensuing deformation fields at the micro-scale would provide highly valuable insight into the relationship between microstructure and macroscopic mechanical response. They would also provide direct validation for numerical simulations that can guide and speed up the design of new materials and microstructures. However, to date, such data have been rare. Here, a one-of-a-kind, multi-modal dataset is presented that combines recent state-of-the-art experimental developments in 3D tomography and high-resolution deformation field measurements.\n\nRelevant Snippet:\n\u2026 maps and indexed as individual features, using the iterative Hough \u2026 Reindexing of the EBSD data with the dictionary indexing \u2026 approach using an evolutionary algorithm. Mater. Charact. \u2026\n\n\n---\n\n## 4. Bayesian Optimization for Diffraction Profile Modeling\nValid Text Names:\n'panchi2023bayesianoptimizationfor pages 1-8', 'panchi2023bayesianoptimizationfor pages 8-13', 'panchi2023bayesianoptimizationfor pages 13-18', 'panchi2023bayesianoptimizationfor pages 18-22', 'panchi2023bayesianoptimizationfor pages 22-26', 'panchi2023bayesianoptimizationfor pages 26-30', 'panchi2023bayesianoptimizationfor pages 30-34', 'panchi2023bayesianoptimizationfor pages 34-39', 'panchi2023bayesianoptimizationfor pages 39-44', 'panchi2023bayesianoptimizationfor pages 44-50', 'panchi2023bayesianoptimizationfor pages 50-54', 'panchi2023bayesianoptimizationfor pages 54-60', 'panchi2023bayesianoptimizationfor pages 60-66', 'panchi2023bayesianoptimizationfor pages 66-70', 'panchi2023bayesianoptimizationfor pages 70-74'\n\nBibTex:\n@article{panchi2023bayesianoptimizationfor,\n author = \"Panchi, Navid\",\n title = \"Bayesian Optimization for Diffraction Profile Modeling\",\n year = \"2023\",\n journal = \"Proceedings of the Institute for Multiscale Simulation\",\n month = \"Jan\",\n doi = \"10.25593/open-fau-1463\",\n url = \"https://doi.org/10.25593/open-fau-1463\",\n publisher = \"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)\"\n}\n\n\n\nUnobtainable Papers:\nDavid J Rowenhorst and William Lenthe. Ebsd pattern center fitting with particle swarm optimization. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.304, doi:10.1093/mam/ozaf048.304.\nYiling Huang, Fan Peng, Xuemei Song, Xingyu Jin, Yuqing Jiang, Wei Zheng, Caifen Jiang, Zhaoqi Wu, and Yi Zeng. Optimisation of ebsd indexing through pattern centre calibration and grain boundary refinement. Journal of microscopy, May 2025. URL: https://doi.org/10.1111/jmi.13434, doi:10.1111/jmi.13434.\nF. Briffod, T. E. J. Edwards, J. Quinta da Fonseca, J. -C. Stinville, D. Texier, and T. Vermeij. Understanding strain localization in metallic materials: a review of high-resolution digital image correlation and related techniques. Science and Technology of Advanced Materials, Mar 2026. URL: https://doi.org/10.1080/14686996.2026.2630488, doi:10.1080/14686996.2026.2630488.\nYunxuan Zhou, Zihao Wang, Wenhui Tao, Yongkang Sun, Junjie Wu, Gang Wang, Yu Xiu, Huiyu Ji, Yulin Liu, Anping Dong, Jie Wang, Jun Wang, Mengmeng Wang, and Qi Liu. Multi-objective optimization design of al-si alloys based on machine learning. Sep 2025. URL: https://doi.org/10.1016/j.calphad.2025.102870, doi:10.1016/j.calphad.2025.102870.\nZihao Ding. Applications of machine learning in crystallographic orientation determination. Text, Jan 2022. URL: https://doi.org/10.1184/r1/20750590, doi:10.1184/r1/20750590.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_9At2TUTAjnJ23z8NGaLTDASo", + "info": { + "start_ts": 397.733964229, + "end_ts": 437.781901546 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring.\nValid Text Names:\n'amini2024sustainabilityinspiredfabrication pages 1-2', 'amini2024sustainabilityinspiredfabrication pages 2-3', 'amini2024sustainabilityinspiredfabrication pages 3-5', 'amini2024sustainabilityinspiredfabrication pages 5-6', 'amini2024sustainabilityinspiredfabrication pages 6-8', 'amini2024sustainabilityinspiredfabrication pages 8-10', 'amini2024sustainabilityinspiredfabrication pages 10-11', 'amini2024sustainabilityinspiredfabrication pages 11-12', 'amini2024sustainabilityinspiredfabrication pages 12-13', 'amini2024sustainabilityinspiredfabrication pages 13-14', 'amini2024sustainabilityinspiredfabrication pages 14-15', 'amini2024sustainabilityinspiredfabrication pages 15-16', 'amini2024sustainabilityinspiredfabrication pages 16-16'\n\nBibTex:\n@article{amini2024sustainabilityinspiredfabrication,\n author = \"Amini, Shahram and Choi, Hongbin and Seche, Wesley and Blagojevic, Alexander and May, Nicholas and Lefler, Benjamin M. and Davis, Skyler L. and Elyahoodayan, Sahar and Tavousi, Pouya and May, Steven J. and Caputo, Gregory A. and Lowe, Terry C. and Hettinger, Jeffrey and Shahbazmohamadi, Sina\",\n title = \"Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring.\",\n year = \"2024\",\n journal = \"Microsystems \\& nanoengineering\",\n volume = \"10 1\",\n pages = \"125\",\n month = \"Sep\",\n doi = \"10.1038/s41378-024-00754-w\",\n url = \"https://doi.org/10.1038/s41378-024-00754-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2055-7434\"\n}\n\n\nAbstract:\nOver the last two decades, platinum group metals (PGMs) and their alloys have dominated as the materials of choice for electrodes in long-term implantable neurostimulation and cardiac rhythm management devices due to their superior conductivity, mechanical and chemical stability, biocompatibility, corrosion resistance, radiopacity, and electrochemical performance. Despite these benefits, PGM manufacturing processes are extremely costly, complex, and challenging with potential health hazards. Additionally, the volatility in PGM prices and their high supply risk, combined with their scarce concentration of approximately 0.01 ppm in the earth's upper crust and limited mining geographical areas, underscores their classification as critical raw materials, thus, their effective recovery or substitution worldwide is of paramount importance. Since postmortem recovery from deceased patients and/or refining of PGMs that are used in the manufacturing of the electrodes and microelectrode arrays is extremely rare, challenging, and highly costly, therefore, substitution of PGM-based electrodes with other biocompatible materials that can yield electrochemical performance values equal or greater than PGMs is the only viable and sustainable solution to reduce and ultimately substitute the use of PGMs in long-term implantable neurostimulation and cardiac rhythm management devices. In this article, we demonstrate for the first time how the novel technique of \"reactive hierarchical surface restructuring\" can be utilized on titanium-that is widely used in many non-stimulation medical device and implant applications-to manufacture biocompatible, low-cost, sustainable, and high-performing neurostimulation and cardiac rhythm management electrodes. We have shown how the surface of titanium electrodes with extremely poor electrochemical performance undergoes compositional and topographical transformations that result in electrodes with outstanding electrochemical performance.\n\nRelevant Snippet:\n\u2026 To achieve the surface quality required for EBSD analysis, ion milling was \u2026 pattern average re-indexing was performed on the dataset. Grain dilation and neighbor confidence index \u2026\n\n\nUnobtainable Papers:\nAli Riza Durmaz, James D. Lamb, Kamilla Zaripova, Madeline Vailhe, Julia T. Pu\u0308rstl, Justine Schulte, Marc Ackermann, McLean P. Echlin, and Tresa M. Pollock. A correlative microscopy dataset for multimodal data fusion and image matching in materials science. ChemRxiv, Apr 2026. URL: https://doi.org/10.26434/chemrxiv.15001724/v1, doi:10.26434/chemrxiv.15001724/v1.\nPatrick T. Brewick, Stuart I. Wright, and David J. Rowenhorst. Nlpar: non-local smoothing for enhanced ebsd pattern indexing. Ultramicroscopy, 200:50-61, May 2019. URL: https://doi.org/10.1016/j.ultramic.2019.02.013, doi:10.1016/j.ultramic.2019.02.013.\nCaleb E. Andrews, Maria Strantza, Nicholas P. Calta, Manyalibo J. Matthews, and Mitra L. Taheri. A denoising autoencoder for improved kikuchi pattern quality and indexing in electron backscatter diffraction. Ultramicroscopy, 253:113810, Nov 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113810, doi:10.1016/j.ultramic.2023.113810.\nA.E. Davis, X. Zeng, R. Thomas, J.R. Kennedy, J. Donoghue, A. Gholinia, P.B. Prangnell, and J. Quinta Da Fonseca. Optimising large-area crystal orientation mapping of nanoscale \u03b2 phase in \u03b1 + \u03b2 titanium alloys using ebsd. Materials Characterization, 194:112371, Dec 2022. URL: https://doi.org/10.1016/j.matchar.2022.112371, doi:10.1016/j.matchar.2022.112371.\nCatherine Zhou. Combinatorial substrate epitaxy and computationally guided epitaxial synthesis investigations of the aemno3 (ae = ca, sr, and ba) polymorphs. Text, Jan 2023. URL: https://doi.org/10.1184/r1/22313545.v1, doi:10.1184/r1/22313545.v1.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_5aBBEXJupU0KmlHg00b79lgu", + "info": { + "start_ts": 397.767522477, + "end_ts": 440.755996379 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nQiwei Shi, Dominique Loisnard, Yang Li, Zhe Chen, Haowei Wang, and St\u00e9phane Roux. Accuracy assessment of crystal orientation indexations by ebsd. Measurement Science and Technology, 35:045030, Jan 2024. URL: https://doi.org/10.1088/1361-6501/ad204d, doi:10.1088/1361-6501/ad204d.\nPatrick Trimby, Mohammed Al-Mosawi, Maisoon Al-Jawad, Stuart Micklethwaite, Zabeada Aslam, Aimo Winkelmann, and Sandra Piazolo. The characterisation of dental enamel using transmission kikuchi diffraction in the scanning electron microscope combined with dynamic template matching. Ultramicroscopy, 260:113940, Jun 2024. URL: https://doi.org/10.1016/j.ultramic.2024.113940, doi:10.1016/j.ultramic.2024.113940.\nFarangis Ram, Stuart Wright, Saransh Singh, and Marc De Graef. Error analysis of the crystal orientations obtained by the dictionary approach to ebsd indexing. Ultramicroscopy, 181:17-26, Oct 2017. URL: https://doi.org/10.1016/j.ultramic.2017.04.016, doi:10.1016/j.ultramic.2017.04.016.\nKatharina Marquardt, Marc De Graef, Saransh Singh, Hauke Marquardt, Anja Rosenthal, and Sanae Koizuimi. Quantitative electron backscatter diffraction (ebsd) data analyses using the dictionary indexing (di) approach: overcoming indexing difficulties on geological materials. American Mineralogist, 102:1843-1855, Sep 2017. URL: https://doi.org/10.2138/am-2017-6062, doi:10.2138/am-2017-6062.\nDavid J. Rowenhorst, Patrick G. Callahan, and H\u00e5kon W. \u00c5nes. Fast radon transforms for high-precision ebsd orientation determination using pyebsdindex. Journal of Applied Crystallography, 57:3-19, Feb 2024. URL: https://doi.org/10.1107/s1600576723010221, doi:10.1107/s1600576723010221.\nK. V. Mani Krishna, R. Madhavan, Mangesh V. Pantawane, Rajarshi Banerjee, and Narendra B. Dahotre. Machine learning based de-noising of electron back scatter patterns of various crystallographic metallic materials fabricated using laser directed energy deposition. Ultramicroscopy, 247:113703, May 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113703, doi:10.1016/j.ultramic.2023.113703.\nM. A. Jackson, E. Pascal, and M. De Graef. Dictionary indexing of electron back-scatter diffraction patterns: a hands-on tutorial. Integrating Materials and Manufacturing Innovation, 8:226-246, May 2019. URL: https://doi.org/10.1007/s40192-019-00137-4, doi:10.1007/s40192-019-00137-4.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_acsPy2saVbrPRcVvQrHXhwqx", + "info": { + "start_ts": 397.799029707, + "end_ts": 471.298576361 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Recent advances and applications of deep learning methods in materials science\nValid Text Names:\n'choudhary2110recentadvancesand pages 1-2', 'choudhary2110recentadvancesand pages 2-3', 'choudhary2110recentadvancesand pages 3-4', 'choudhary2110recentadvancesand pages 3-3', 'choudhary2110recentadvancesand pages 4-5', 'choudhary2110recentadvancesand pages 5-6', 'choudhary2110recentadvancesand pages 6-8', 'choudhary2110recentadvancesand pages 8-10', 'choudhary2110recentadvancesand pages 8-8', 'choudhary2110recentadvancesand pages 10-11', 'choudhary2110recentadvancesand pages 11-12', 'choudhary2110recentadvancesand pages 11-11', 'choudhary2110recentadvancesand pages 14-15', 'choudhary2110recentadvancesand pages 16-17', 'choudhary2110recentadvancesand pages 17-18', 'choudhary2110recentadvancesand pages 18-19', 'choudhary2110recentadvancesand pages 19-20', 'choudhary2110recentadvancesand pages 20-21', 'choudhary2110recentadvancesand pages 20-20', 'choudhary2110recentadvancesand pages 21-22', 'choudhary2110recentadvancesand pages 22-23', 'choudhary2110recentadvancesand pages 22-22', 'choudhary2110recentadvancesand pages 23-24', 'choudhary2110recentadvancesand pages 24-24', 'choudhary2110recentadvancesand pages 25-26'\n\nBibTex:\n@article{choudhary2022recentadvancesand,\n author = \"Choudhary, Kamal and DeCost, Brian and Chen, Chi and Jain, Anubhav and Tavazza, Francesca and Cohn, Ryan and Park, Cheol Woo and Choudhary, Alok and Agrawal, Ankit and Billinge, Simon J. L. and Holm, Elizabeth and Ong, Shyue Ping and Wolverton, Chris\",\n title = \"Recent advances and applications of deep learning methods in materials science\",\n year = \"2022\",\n journal = \"npj Computational Materials\",\n volume = \"8\",\n month = \"Apr\",\n doi = \"10.1038/s41524-022-00734-6\",\n url = \"https://doi.org/10.1038/s41524-022-00734-6\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\",\n pages = \"1-26\"\n}\n\n\nAbstract:\nAbstractDeep learning (DL) is one of the fastest-growing topics in materials data science, with rapidly emerging applications spanning atomistic, image-based, spectral, and textual data modalities. DL allows analysis of unstructured data and automated identification of features. The recent development of large materials databases has fueled the application of DL methods in atomistic prediction in particular. In contrast, advances in image and spectral data have largely leveraged synthetic data enabled by high-quality forward models as well as by generative unsupervised DL methods. In this article, we present a high-level overview of deep learning methods followed by a detailed discussion of recent developments of deep learning in atomistic simulation, materials imaging, spectral analysis, and natural language processing. For each modality we discuss applications involving both theoretical and experimental data, typical modeling approaches with their strengths and limitations, and relevant publicly available software and datasets. We conclude the review with a discussion of recent cross-cutting work related to uncertainty quantification in this field and a brief perspective on limitations, challenges, and potential growth areas for DL methods in materials science.\n\nRelevant Snippet:\n\u2026 EBSD indexing capable of taking an EBSD image as input and predicting the three Euler angles representing the orientation that would have led to the given EBSD \u2026 post-processing of \u2026\n\n\nUnobtainable Papers:\nGrace Guinan, Addison Salvador, Michelle A. Smeaton, Andrew Glaws, Hilary Egan, Brian C. Wyatt, Babak Anasori, Kevin R. Fiedler, Matthew J. Olszta, and Steven R. Spurgeon. Mind the gap: bridging the divide between ai aspirations and the reality of autonomous microscopy. Jun 2025. URL: https://doi.org/10.1063/5.0267699, doi:10.1063/5.0267699.\nMohammad Mazedul Islam, T. M. A. Iqbal Bin Belal, Md. Arif Hossain Chowdhury Anik, and Ahmed Sharif. Predicting materials using variational autoencoders: a systematic literature review. Archives of Computational Methods in Engineering, 33(5):6599-6650, Jan 2026. URL: https://doi.org/10.1007/s11831-025-10485-4, doi:10.1007/s11831-025-10485-4.\nChunyu Guo, Xiangyu Tang, Yu\u2019e Chen, Changyou Gao, Qinglin Shan, Heyi Wei, Xusheng Liu, Chuncheng Lu, Meixia Fu, Enhui Wang, Xinhong Liu, Xinmei Hou, and Yanglong Hou. From microstructure to performance optimization: innovative applications of computer vision in materials science. International Journal of Minerals, Metallurgy and Materials, 33(1):94-115, Dec 2026. URL: https://doi.org/10.1007/s12613-025-3217-4, doi:10.1007/s12613-025-3217-4.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_SvSRR7s0yK3PpnhQQ0Vw5Xl4", + "info": { + "start_ts": 397.833985678, + "end_ts": 435.362992784 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\nValid Text Names:\n'polonsky2019threedimensionalanalysisand pages 1-2', 'polonsky2019threedimensionalanalysisand pages 2-5', 'polonsky2019threedimensionalanalysisand pages 5-6', 'polonsky2019threedimensionalanalysisand pages 6-8', 'polonsky2019threedimensionalanalysisand pages 8-11', 'polonsky2019threedimensionalanalysisand pages 11-13', 'polonsky2019threedimensionalanalysisand pages 13-14', 'polonsky2019threedimensionalanalysisand pages 14-15', 'polonsky2019threedimensionalanalysisand pages 15-15'\n\nBibTex:\n@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a cloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline variables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore, a comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using BisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects for impact in material science are also discussed.\n\nRelevant Snippet:\n\u2026 about the grain segmentation misorientation tolerance in \u2026 Automated orientation imaging analysis via EBSD first \u2026 During the reconstruction process, various cleanup steps are employed \u2026\n\n\n---\n\n## 2. 3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry\nValid Text Names:\n'sparks20243dreconstructionof pages 1-2', 'sparks20243dreconstructionof pages 2-6', 'sparks20243dreconstructionof pages 6-7', 'sparks20243dreconstructionof pages 7-9', 'sparks20243dreconstructionof pages 9-12', 'sparks20243dreconstructionof pages 12-14', 'sparks20243dreconstructionof pages 14-16', 'sparks20243dreconstructionof pages 16-18', 'sparks20243dreconstructionof pages 18-19', 'sparks20243dreconstructionof pages 19-21', 'sparks20243dreconstructionof pages 21-23', 'sparks20243dreconstructionof pages 23-24', 'sparks20243dreconstructionof pages 24-26', 'sparks20243dreconstructionof pages 26-28', 'sparks20243dreconstructionof pages 28-30', 'sparks20243dreconstructionof pages 30-31', 'sparks20243dreconstructionof pages 31-31'\n\nBibTex:\n@article{sparks20243dreconstructionof,\n author = \"Sparks, Gregory and Mason, Simon A. and Chapman, Michael G. and Park, Jun-Sang and Sharma, Hemant and Kenesei, Peter and Niezgoda, Stephen R. and Mills, Michael J. and Uchic, Michael D. and Shade, Paul A. and Obstalecki, Mark\",\n title = \"3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jul\",\n doi = \"10.1007/s40192-024-00370-6\",\n url = \"https://doi.org/10.1007/s40192-024-00370-6\",\n volume = \"13\",\n pages = \"773-803\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"3\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 EBSD, pattern center fitting for every section, automated \u2026 Grains were segmented using a misorientation threshold of 3 \u25e6\u2026 3D pipelines that were used for cleanup and segmentation, \u2026\n\n\n---\n\n## 3. Recent Progress of Digital Reconstruction in Polycrystalline Materials\nValid Text Names:\n'chen2025recentprogressof pages 1-2', 'chen2025recentprogressof pages 2-4', 'chen2025recentprogressof pages 4-7', 'chen2025recentprogressof pages 10-12', 'chen2025recentprogressof pages 12-15', 'chen2025recentprogressof pages 15-18', 'chen2025recentprogressof pages 18-19', 'chen2025recentprogressof pages 19-20', 'chen2025recentprogressof pages 20-24', 'chen2025recentprogressof pages 24-25', 'chen2025recentprogressof pages 25-27', 'chen2025recentprogressof pages 27-28', 'chen2025recentprogressof pages 28-32', 'chen2025recentprogressof pages 32-34', 'chen2025recentprogressof pages 34-36', 'chen2025recentprogressof pages 38-40', 'chen2025recentprogressof pages 41-42', 'chen2025recentprogressof pages 42-43', 'chen2025recentprogressof pages 43-44', 'chen2025recentprogressof pages 45-46', 'chen2025recentprogressof pages 46-47', 'chen2025recentprogressof pages 47-48', 'chen2025recentprogressof pages 48-49', 'chen2025recentprogressof pages 49-50', 'chen2025recentprogressof pages 50-51'\n\nBibTex:\n@article{chen2025recentprogressof,\n author = \"Chen, Bingbing and Li, Dongfeng and Davies, Peter and Johnston, Richard and Ge, Xiangyun and Li, Chenfeng\",\n title = \"Recent Progress of Digital Reconstruction in Polycrystalline Materials\",\n year = \"2025\",\n journal = \"Archives of Computational Methods in Engineering\",\n month = \"Mar\",\n doi = \"10.1007/s11831-025-10245-4\",\n url = \"https://doi.org/10.1007/s11831-025-10245-4\"\n}\n\n\nAbstract:\n\n This study comprehensively reviews recent advances in the digital reconstruction of polycrystalline materials. Digital reconstruction serves as both a representative volume element for multiscale modelling and a source of quantitative data for microstructure characterisation. Three main types of digital reconstruction in polycrystalline materials exist: (i) experimental reconstruction, which links processing-structure-properties-performance by reconstructing actual polycrystalline microstructures using destructive or non-destructive methods; (ii) physics-based models, which replicate evolutionary processes to establish processing-structure linkages, including cellular automata, Monte Carlo, vertex/front tracking, level set, machine learning, and phase field methods; and (iii) geometry-based models, which create ensembles of statistically equivalent polycrystalline microstructures for structure-properties-performance linkages, using simplistic morphology, Voronoi tessellation, ellipsoid packing, texture synthesis, high-order, reduced-order, and machine learning methods. This work reviews the key features, procedures, advantages, and limitations of these methods, with a particular focus on their application in constructing processing-structure-properties-performance linkages. Finally, it summarises the conclusions, challenges, and future directions for digital reconstruction in polycrystalline materials within the framework of computational materials engineering.\n\nRelevant Snippet:\n\u2026 Several automated devices are commonly used for serial \u2026 data mapping for identifying grains and twins in EBSD (Fig 3c). For \u2026 faster than GFIB [33] and subsequent surface cleanup at \u2026\n\n\nUnobtainable Papers:\nJiwoong Kang. Kinetics and connectivity of grain boundaries in three-dimensional metallic systems. Text, Jan 2022. URL: https://doi.org/10.7302/6025, doi:10.7302/6025.\nMark Taylor, Albert D Smith, Jack M Donoghue, Timothy L Burnett, and Ed Pickering. In-situ heating-stage ebsd validation of algorithms for prior-austenite grain reconstruction in steel. Scripta Materialia, Mar 2024. URL: https://doi.org/10.1016/j.scriptamat.2023.115924, doi:10.1016/j.scriptamat.2023.115924.\nAlistair Garner, Jack Donoghue, Remco Geurts, Yasser Al Aboura, Bart\u0142omiej Winiarski, Phil B. Prangnell, and Tim L. Burnett. Large-scale serial sectioning of environmentally assisted cracks in 7xxx al alloys using femtosecond laser-pfib. Materials Characterization, 188:111890, Jun 2022. URL: https://doi.org/10.1016/j.matchar.2022.111890, doi:10.1016/j.matchar.2022.111890.\nH\u00e5kon W. \u00c5nes, Antonius T.J. van Helvoort, and Knut Marthinsen. Orientation dependent pinning of (sub)grains by dispersoids during recovery and recrystallization in an al\u2013mn alloy. Acta Materialia, 248:118761, Apr 2023. URL: https://doi.org/10.1016/j.actamat.2023.118761, doi:10.1016/j.actamat.2023.118761.\nH. Latuner, Julie Favre, A. Helstroffer, G. Inacio Da Rosa, P. Joly, C. Maurice, E. Plancher, M. Roth, C. Desrayaud, and G. Kermouche. Identification of recrystallized grains in ebsd maps: automated adaptive thresholding using a gaussian mixture approach. Materials Characterization, 231:115920, Jan 2026. URL: https://doi.org/10.1016/j.matchar.2025.115920, doi:10.1016/j.matchar.2025.115920.\nThomas Girerd, Giovanna Mart\u00ednez-Arellano, Adam Thomas Clare, and Alistair Speidel. A facile methodology to identify microstructural grains on etched surfaces using panoptic segmentation. Sep 2025. URL: https://doi.org/10.1016/j.matchar.2025.115224, doi:10.1016/j.matchar.2025.115224.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_yBdeyyzRACDmvHLG4V5YZoxC", + "info": { + "start_ts": 397.873479618, + "end_ts": 440.096785711 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional patents. This may be because nothing matched the search query, or because all found patents have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=0 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "patent_search", + "tool_call_id": "call_4uZ0w5ewQsn4Ra9JRMzH48rh", + "info": { + "start_ts": 397.913352854, + "end_ts": 399.574481165 + } + }, + { + "role": "tool", + "content": "Retrieved Patents:\n## 1. Optical package for a lidar sensor system and lidar sensor system technical \u2026\nValid Text Names:\n'US20220276352A1 pages 10-18', 'US20220276352A1 pages 24-25', 'US20220276352A1 pages 25-26', 'US20220276352A1 pages 26-27', 'US20220276352A1 pages 27-28', 'US20220276352A1 pages 29-30', 'US20220276352A1 pages 30-31', 'US20220276352A1 pages 31-32', 'US20220276352A1 pages 32-33', 'US20220276352A1 pages 34-35', 'US20220276352A1 pages 37-38', 'US20220276352A1 pages 38-39', 'US20220276352A1 pages 39-40', 'US20220276352A1 pages 41-42', 'US20220276352A1 pages 42-43', 'US20220276352A1 pages 44-45', 'US20220276352A1 pages 46-46', 'US20220276352A1 pages 47-48', 'US20220276352A1 pages 48-49', 'US20220276352A1 pages 49-50', 'US20220276352A1 pages 52-53', 'US20220276352A1 pages 58-59', 'US20220276352A1 pages 59-60', 'US20220276352A1 pages 60-61', 'US20220276352A1 pages 61-62'\n\nBibTex:\n@article{US20220276352A1,\n author = \"Ferreira, Ricardo\",\n title = \"Optical package for a lidar sensor system and lidar sensor system technical \u2026\",\n year = \"2022\",\n journal = \"Patent (WO,EP,US,CN,CA,DE)\"\n}\n\n\nRelevant Snippet:\nThe present disclosure relates to various embodiments of an optical package for a LIDAR sensor system having a substrate including an array of a plurality of capacitors formed in the substrate, a plurality of switches formed in the substrate, where each switch is connected between at least one \u2026\n\n\n---\n\n## 2. Hyperspectral sensing system and methods\nValid Text Names:\n'US11650190B2 pages 1-4', 'US11650190B2 pages 4-6', 'US11650190B2 pages 6-9', 'US11650190B2 pages 12-15', 'US11650190B2 pages 18-21', 'US11650190B2 pages 26-29', 'US11650190B2 pages 29-31', 'US11650190B2 pages 31-34', 'US11650190B2 pages 34-37', 'US11650190B2 pages 39-41', 'US11650190B2 pages 41-44', 'US11650190B2 pages 46-49', 'US11650190B2 pages 53-56', 'US11650190B2 pages 58-59', 'US11650190B2 pages 61-63', 'US11650190B2 pages 65-67', 'US11650190B2 pages 69-71', 'US11650190B2 pages 81-83', 'US11650190B2 pages 87-89', 'US11650190B2 pages 89-91', 'US11650190B2 pages 96-98', 'US11650190B2 pages 100-102', 'US11650190B2 pages 104-106', 'US11650190B2 pages 106-108', 'US11650190B2 pages 110-112'\n\nBibTex:\n@article{US11650190B2,\n author = \"Tufillaro, Nicholas\",\n title = \"Hyperspectral sensing system and methods\",\n year = \"2023\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nA method for retrieving a corrected spectrum from a measured spectrum (e.g., retrieving a top-of-water spectrum from a measured top-of-atmosphere spectrum) includes creating a scene-specific model of a region of interest and performing a ray-tracing simulation to simulate rays of light that would \u2026\n\n\n---\n\n## 3. Multidimensional materials sensing systems and methods\nValid Text Names:\n'US12517250B2 pages 1-4', 'US12517250B2 pages 4-6', 'US12517250B2 pages 6-9', 'US12517250B2 pages 9-11', 'US12517250B2 pages 11-14', 'US12517250B2 pages 14-17', 'US12517250B2 pages 19-22', 'US12517250B2 pages 22-25', 'US12517250B2 pages 33-35', 'US12517250B2 pages 38-41', 'US12517250B2 pages 41-43', 'US12517250B2 pages 43-46', 'US12517250B2 pages 46-48', 'US12517250B2 pages 48-51', 'US12517250B2 pages 51-53', 'US12517250B2 pages 53-55', 'US12517250B2 pages 55-57', 'US12517250B2 pages 59-61', 'US12517250B2 pages 61-62', 'US12517250B2 pages 64-66', 'US12517250B2 pages 66-68', 'US12517250B2 pages 68-70', 'US12517250B2 pages 74-76', 'US12517250B2 pages 78-80', 'US12517250B2 pages 86-86'\n\nBibTex:\n@article{US12517250B2,\n author = \"Kotov, Nicholas A.\",\n title = \"Multidimensional materials sensing systems and methods\",\n year = \"2026\",\n journal = \"Patent (WO,EP,US,CN,JP,KR)\"\n}\n\n\nRelevant Snippet:\nMultidimensional Light Imaging, Detection, and Ranging (LIDAR) systems and methods optionally include a laser device configured to generate a plurality of light pulses emitted towards an object, a detector configured to receive a portion of the plurality of light pulses returned from the object, \u2026\n\n\n---\n\n## 4. Super system on chip\nValid Text Names:\n'US12380059B2 pages 1-4', 'US12380059B2 pages 18-20', 'US12380059B2 pages 20-23', 'US12380059B2 pages 23-26', 'US12380059B2 pages 29-32', 'US12380059B2 pages 37-40', 'US12380059B2 pages 43-46', 'US12380059B2 pages 59-62', 'US12380059B2 pages 75-78', 'US12380059B2 pages 86-89', 'US12380059B2 pages 89-92', 'US12380059B2 pages 100-103', 'US12380059B2 pages 111-114', 'US12380059B2 pages 125-128', 'US12380059B2 pages 137-140', 'US12380059B2 pages 148-150', 'US12380059B2 pages 154-156', 'US12380059B2 pages 156-158', 'US12380059B2 pages 185-187', 'US12380059B2 pages 199-201', 'US12380059B2 pages 203-205', 'US12380059B2 pages 207-210', 'US12380059B2 pages 214-216', 'US12380059B2 pages 218-220', 'US12380059B2 pages 239-241'\n\nBibTex:\n@article{US12380059B2,\n author = \"Mazed, Mohammad A.\",\n title = \"Super system on chip\",\n year = \"2025\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nA Super System is disclosed and its inputs/outputs are coupled with a Mach-Zehnder interferometer (MZI), wherein the Mach-Zehnder interferometer (MZI) can be coupled with a first optical waveguide either in a two-dimensional (2-D) or in a three-dimensional (3-D) arrangement. The first optical \u2026\n\n\n---\n\n## 5. Super system on chip\nValid Text Names:\n'US11320588B1 pages 1-4', 'US11320588B1 pages 4-7', 'US11320588B1 pages 7-11', 'US11320588B1 pages 11-14', 'US11320588B1 pages 17-20', 'US11320588B1 pages 22-25', 'US11320588B1 pages 25-28', 'US11320588B1 pages 31-34', 'US11320588B1 pages 36-39', 'US11320588B1 pages 48-50', 'US11320588B1 pages 50-53', 'US11320588B1 pages 59-62', 'US11320588B1 pages 65-67', 'US11320588B1 pages 67-70', 'US11320588B1 pages 85-87', 'US11320588B1 pages 92-95', 'US11320588B1 pages 95-98', 'US11320588B1 pages 98-100', 'US11320588B1 pages 103-106', 'US11320588B1 pages 106-109', 'US11320588B1 pages 109-112', 'US11320588B1 pages 115-117', 'US11320588B1 pages 117-120', 'US11320588B1 pages 120-122', 'US11320588B1 pages 197-200'\n\nBibTex:\n@article{US11320588B1,\n author = \"Mazed, Mohammad A.\",\n title = \"Super system on chip\",\n year = \"2022\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nA Super System on Chip (SSoC) coupled with a photonic neural learning processor (PNLP), one or more quantum bits (qubits) and a machine learning algorithm for ultrafast data processing, image processing/recognition, deep learning/meta-learning and self-learning is disclosed. The Super System on \u2026\n\n\n---\n\n## 6. Methods and apparatus for three-dimensional (3D) imaging\nValid Text Names:\n'US11112503B2 pages 1-2', 'US11112503B2 pages 2-5', 'US11112503B2 pages 5-12', 'US11112503B2 pages 12-15', 'US11112503B2 pages 15-19', 'US11112503B2 pages 19-20', 'US11112503B2 pages 20-21', 'US11112503B2 pages 21-22', 'US11112503B2 pages 22-23', 'US11112503B2 pages 23-24', 'US11112503B2 pages 24-25', 'US11112503B2 pages 25-25', 'US11112503B2 pages 25-26', 'US11112503B2 pages 26-27', 'US11112503B2 pages 27-28', 'US11112503B2 pages 28-29', 'US11112503B2 pages 29-30', 'US11112503B2 pages 30-31', 'US11112503B2 pages 31-32', 'US11112503B2 pages 32-33', 'US11112503B2 pages 33-34', 'US11112503B2 pages 34-35', 'US11112503B2 pages 35-36', 'US11112503B2 pages 36-37', 'US11112503B2 pages 37-38'\n\nBibTex:\n@article{US11112503B2,\n author = \"Fried, Dale G.\",\n title = \"Methods and apparatus for three-dimensional (3D) imaging\",\n year = \"2021\",\n journal = \"Patent (WO,EP,US,CA)\"\n}\n\n\nRelevant Snippet:\nA method of imaging a scene includes estimating multiple three-dimensional (3D) representations, each of which corresponds to a respective portion of the scene. Neighboring portions of the scene area are at least partially overlapping. Each 3D representation is estimated by illuminating the \u2026\n\n\n---\n\n## 7. Smart device\nValid Text Names:\n'US10046229B2 pages 1-2', 'US10046229B2 pages 2-4', 'US10046229B2 pages 13-17', 'US10046229B2 pages 17-18', 'US10046229B2 pages 20-21', 'US10046229B2 pages 21-22', 'US10046229B2 pages 23-24', 'US10046229B2 pages 24-25', 'US10046229B2 pages 28-29', 'US10046229B2 pages 29-30', 'US10046229B2 pages 30-30', 'US10046229B2 pages 31-32', 'US10046229B2 pages 34-35', 'US10046229B2 pages 36-37', 'US10046229B2 pages 38-39', 'US10046229B2 pages 40-41', 'US10046229B2 pages 44-45', 'US10046229B2 pages 47-48', 'US10046229B2 pages 49-50', 'US10046229B2 pages 50-51', 'US10046229B2 pages 53-54', 'US10046229B2 pages 54-55', 'US10046229B2 pages 55-56', 'US10046229B2 pages 60-61', 'US10046229B2 pages 62-63'\n\nBibTex:\n@article{US10046229B2,\n author = \"Tran, Bao\",\n title = \"Smart device\",\n year = \"2018\",\n journal = \"Patent (US)\"\n}\n\n\nRelevant Snippet:\nAn Internet of Thing (IoT) device includes a body with a processor, a camera and a wireless transceiver coupled to the processor.\n\n\nUnobtainable Patents:\nJiwoong Kang. Kinetics and connectivity of grain boundaries in three-dimensional metallic systems. Text, Jan 2022. URL: https://doi.org/10.7302/6025, doi:10.7302/6025.\nQiwei Shi, Dominique Loisnard, Yang Li, Zhe Chen, Haowei Wang, and St\u00e9phane Roux. Accuracy assessment of crystal orientation indexations by ebsd. Measurement Science and Technology, 35:045030, Jan 2024. URL: https://doi.org/10.1088/1361-6501/ad204d, doi:10.1088/1361-6501/ad204d.\nPatrick Trimby, Mohammed Al-Mosawi, Maisoon Al-Jawad, Stuart Micklethwaite, Zabeada Aslam, Aimo Winkelmann, and Sandra Piazolo. The characterisation of dental enamel using transmission kikuchi diffraction in the scanning electron microscope combined with dynamic template matching. Ultramicroscopy, 260:113940, Jun 2024. URL: https://doi.org/10.1016/j.ultramic.2024.113940, doi:10.1016/j.ultramic.2024.113940.\nYiling Huang, Fan Peng, Xuemei Song, Xingyu Jin, Yuqing Jiang, Wei Zheng, Caifen Jiang, Zhaoqi Wu, and Yi Zeng. Optimisation of ebsd indexing through pattern centre calibration and grain boundary refinement. Journal of microscopy, May 2025. URL: https://doi.org/10.1111/jmi.13434, doi:10.1111/jmi.13434.\nDavid J. Rowenhorst, Patrick G. Callahan, and H\u00e5kon W. \u00c5nes. Fast radon transforms for high-precision ebsd orientation determination using pyebsdindex. Journal of Applied Crystallography, 57:3-19, Feb 2024. URL: https://doi.org/10.1107/s1600576723010221, doi:10.1107/s1600576723010221.\nGrace Guinan, Addison Salvador, Michelle A. Smeaton, Andrew Glaws, Hilary Egan, Brian C. Wyatt, Babak Anasori, Kevin R. Fiedler, Matthew J. Olszta, and Steven R. Spurgeon. Mind the gap: bridging the divide between ai aspirations and the reality of autonomous microscopy. Jun 2025. URL: https://doi.org/10.1063/5.0267699, doi:10.1063/5.0267699.\nMark Taylor, Albert D Smith, Jack M Donoghue, Timothy L Burnett, and Ed Pickering. In-situ heating-stage ebsd validation of algorithms for prior-austenite grain reconstruction in steel. Scripta Materialia, Mar 2024. URL: https://doi.org/10.1016/j.scriptamat.2023.115924, doi:10.1016/j.scriptamat.2023.115924.\nCaleb E. Andrews, Maria Strantza, Nicholas P. Calta, Manyalibo J. Matthews, and Mitra L. Taheri. A denoising autoencoder for improved kikuchi pattern quality and indexing in electron backscatter diffraction. Ultramicroscopy, 253:113810, Nov 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113810, doi:10.1016/j.ultramic.2023.113810.\nA.E. Davis, X. Zeng, R. Thomas, J.R. Kennedy, J. Donoghue, A. Gholinia, P.B. Prangnell, and J. Quinta Da Fonseca. Optimising large-area crystal orientation mapping of nanoscale \u03b2 phase in \u03b1 + \u03b2 titanium alloys using ebsd. Materials Characterization, 194:112371, Dec 2022. URL: https://doi.org/10.1016/j.matchar.2022.112371, doi:10.1016/j.matchar.2022.112371.\nThomas Girerd, Giovanna Mart\u00ednez-Arellano, Adam Thomas Clare, and Alistair Speidel. A facile methodology to identify microstructural grains on etched surfaces using panoptic segmentation. Sep 2025. URL: https://doi.org/10.1016/j.matchar.2025.115224, doi:10.1016/j.matchar.2025.115224.\nF. Briffod, T. E. J. Edwards, J. Quinta da Fonseca, J. -C. Stinville, D. Texier, and T. Vermeij. Understanding strain localization in metallic materials: a review of high-resolution digital image correlation and related techniques. Science and Technology of Advanced Materials, Mar 2026. URL: https://doi.org/10.1080/14686996.2026.2630488, doi:10.1080/14686996.2026.2630488.\nH. Latuner, Julie Favre, A. Helstroffer, G. Inacio Da Rosa, P. Joly, C. Maurice, E. Plancher, M. Roth, C. Desrayaud, and G. Kermouche. Identification of recrystallized grains in ebsd maps: automated adaptive thresholding using a gaussian mixture approach. Materials Characterization, 231:115920, Jan 2026. URL: https://doi.org/10.1016/j.matchar.2025.115920, doi:10.1016/j.matchar.2025.115920.\nCatherine Zhou. Combinatorial substrate epitaxy and computationally guided epitaxial synthesis investigations of the aemno3 (ae = ca, sr, and ba) polymorphs. Text, Jan 2023. URL: https://doi.org/10.1184/r1/22313545.v1, doi:10.1184/r1/22313545.v1.\nDavid J Rowenhorst and William Lenthe. Ebsd pattern center fitting with particle swarm optimization. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.304, doi:10.1093/mam/ozaf048.304.\nAli Riza Durmaz, James D. Lamb, Kamilla Zaripova, Madeline Vailhe, Julia T. Pu\u0308rstl, Justine Schulte, Marc Ackermann, McLean P. Echlin, and Tresa M. Pollock. A correlative microscopy dataset for multimodal data fusion and image matching in materials science. ChemRxiv, Apr 2026. URL: https://doi.org/10.26434/chemrxiv.15001724/v1, doi:10.26434/chemrxiv.15001724/v1.\nKatharina Marquardt, Marc De Graef, Saransh Singh, Hauke Marquardt, Anja Rosenthal, and Sanae Koizuimi. Quantitative electron backscatter diffraction (ebsd) data analyses using the dictionary indexing (di) approach: overcoming indexing difficulties on geological materials. American Mineralogist, 102:1843-1855, Sep 2017. URL: https://doi.org/10.2138/am-2017-6062, doi:10.2138/am-2017-6062.\nFarangis Ram, Stuart Wright, Saransh Singh, and Marc De Graef. Error analysis of the crystal orientations obtained by the dictionary approach to ebsd indexing. Ultramicroscopy, 181:17-26, Oct 2017. URL: https://doi.org/10.1016/j.ultramic.2017.04.016, doi:10.1016/j.ultramic.2017.04.016.\nPatrick T. Brewick, Stuart I. Wright, and David J. Rowenhorst. Nlpar: non-local smoothing for enhanced ebsd pattern indexing. Ultramicroscopy, 200:50-61, May 2019. URL: https://doi.org/10.1016/j.ultramic.2019.02.013, doi:10.1016/j.ultramic.2019.02.013.\nZihao Ding. Applications of machine learning in crystallographic orientation determination. Text, Jan 2022. URL: https://doi.org/10.1184/r1/20750590, doi:10.1184/r1/20750590.\nK. V. Mani Krishna, R. Madhavan, Mangesh V. Pantawane, Rajarshi Banerjee, and Narendra B. Dahotre. Machine learning based de-noising of electron back scatter patterns of various crystallographic metallic materials fabricated using laser directed energy deposition. Ultramicroscopy, 247:113703, May 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113703, doi:10.1016/j.ultramic.2023.113703.\nAlistair Garner, Jack Donoghue, Remco Geurts, Yasser Al Aboura, Bart\u0142omiej Winiarski, Phil B. Prangnell, and Tim L. Burnett. Large-scale serial sectioning of environmentally assisted cracks in 7xxx al alloys using femtosecond laser-pfib. Materials Characterization, 188:111890, Jun 2022. URL: https://doi.org/10.1016/j.matchar.2022.111890, doi:10.1016/j.matchar.2022.111890.\nMohammad Mazedul Islam, T. M. A. Iqbal Bin Belal, Md. Arif Hossain Chowdhury Anik, and Ahmed Sharif. Predicting materials using variational autoencoders: a systematic literature review. Archives of Computational Methods in Engineering, 33(5):6599-6650, Jan 2026. URL: https://doi.org/10.1007/s11831-025-10485-4, doi:10.1007/s11831-025-10485-4.\nYunxuan Zhou, Zihao Wang, Wenhui Tao, Yongkang Sun, Junjie Wu, Gang Wang, Yu Xiu, Huiyu Ji, Yulin Liu, Anping Dong, Jie Wang, Jun Wang, Mengmeng Wang, and Qi Liu. Multi-objective optimization design of al-si alloys based on machine learning. Sep 2025. URL: https://doi.org/10.1016/j.calphad.2025.102870, doi:10.1016/j.calphad.2025.102870.\nM. A. Jackson, E. Pascal, and M. De Graef. Dictionary indexing of electron back-scatter diffraction patterns: a hands-on tutorial. Integrating Materials and Manufacturing Innovation, 8:226-246, May 2019. URL: https://doi.org/10.1007/s40192-019-00137-4, doi:10.1007/s40192-019-00137-4.\nH\u00e5kon W. \u00c5nes, Antonius T.J. van Helvoort, and Knut Marthinsen. Orientation dependent pinning of (sub)grains by dispersoids during recovery and recrystallization in an al\u2013mn alloy. Acta Materialia, 248:118761, Apr 2023. URL: https://doi.org/10.1016/j.actamat.2023.118761, doi:10.1016/j.actamat.2023.118761.\nChunyu Guo, Xiangyu Tang, Yu\u2019e Chen, Changyou Gao, Qinglin Shan, Heyi Wei, Xusheng Liu, Chuncheng Lu, Meixia Fu, Enhui Wang, Xinhong Liu, Xinmei Hou, and Yanglong Hou. From microstructure to performance optimization: innovative applications of computer vision in materials science. International Journal of Minerals, Metallurgy and Materials, 33(1):94-115, Dec 2026. URL: https://doi.org/10.1007/s12613-025-3217-4, doi:10.1007/s12613-025-3217-4.\n\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=16 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "patent_search", + "tool_call_id": "call_WJzVeGxOMudv7uXJmeYdSe3F", + "info": { + "start_ts": 397.955491698, + "end_ts": 607.461565427 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_uh7IvP3yQ85eCVKwBxCbjRkg", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD \\\"evolutionary algorithm\\\" indexing pattern matching\"}", + "name": "paper_search" + } + }, + { + "id": "call_PGusrK3loi2SP0YX9X3dJxyb", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD \\\"parameter optimization\\\" Hough transform indexing\"}", + "name": "paper_search" + } + }, + { + "id": "call_KI8Kw21UgTxchIJAACgrtWy6", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD pattern center optimization differential evolution particle swarm\"}", + "name": "paper_search" + } + }, + { + "id": "call_AFeXRuxf8pxFGq9DkB0yS3lq", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD dictionary indexing low confidence Hough reindexing\"}", + "name": "paper_search" + } + }, + { + "id": "call_rjeQSC8BesdXWTotR8LKnLxa", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated acquisition dwell time beam current optimization pattern quality\"}", + "name": "paper_search" + } + }, + { + "id": "call_x3Q5do7NWtrRZiM5mvwykX6y", + "type": "function", + "function": { + "arguments": "{\"query\": \"DREAM.3D parameter sweep grain segmentation misorientation tolerance BisQue EBSD\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Effects of SEM Distortions on EBSD Grain Size Measurements\nValid Text Names:\n'timberlake2026effectsofsem pages 1-2', 'timberlake2026effectsofsem pages 2-4', 'timberlake2026effectsofsem pages 4-6', 'timberlake2026effectsofsem pages 6-8', 'timberlake2026effectsofsem pages 8-14', 'timberlake2026effectsofsem pages 14-17', 'timberlake2026effectsofsem pages 17-17'\n\nBibTex:\n@article{timberlake2026effectsofsem,\n author = \"Timberlake, D. M. and Payton, E. J.\",\n title = \"Effects of SEM Distortions on EBSD Grain Size Measurements\",\n year = \"2026\",\n journal = \"Metallography, Microstructure, and Analysis\",\n volume = \"15\",\n pages = \"489-505\",\n month = \"Feb\",\n doi = \"10.1007/s13632-026-01317-3\",\n url = \"https://doi.org/10.1007/s13632-026-01317-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2192-9262\"\n}\n\n\nAbstract:\nAbstract\n Systematic discrepancies persist between results from grain size measurement standards, such as ASTM E 112 and ASTM E 2627, when different imaging techniques and analysis methods are used. The source of these discrepancies may be differences in imaging physics, resolution limits, grain boundary definitions, post-processing practices, or combinations thereof. Geometric distortions and drift effects in scanning electron microscope imaging are systematically modeled using synthetic microstructures to assess their impact on electron backscatter diffraction-based grain size measurement accuracy. Standard methods are compared to a lesser-known method, the Triple Point Count (TPC). Most distortion types introduce smaller biases than the expected accuracy of the methods under typical imaging conditions. TPC provides consistent performance across varying resolutions and sampling conditions. This provides further evidence that reported discrepancies are largely driven by an assumption made in the estimate for converting intercept measurements to ASTM grain size numbers, as recently reported by Evans et al.\n\nRelevant Snippet:\n\u2026 are re-gridded to match the same pixel spacing as the original synthetic EBSD data and any \u2026 are included to reduce error from poorly indexed or partially resolved grains. The mean \u2026\n\n\nUnobtainable Papers:\nWenyuan Zhao, Qingguo Ren, Zhihao Yao, Jie Zhao, He Jiang, and Jianxin Dong. Multi-scale calculation-aided composition optimization design of low-density and high-strength nickel-based superalloy. Jul 2023. URL: https://doi.org/10.1007/s11661-023-07133-6, doi:10.1007/s11661-023-07133-6.\nLaura Mill\u00e1n, Gabriel Kronberger, J. Ignacio Hidalgo, Ricardo Fern\u00e1ndez, Oscar Garnica, and Gaspar Gonz\u00e1lez-Doncel. Estimation of grain-level residual stresses in a quenched cylindrical sample of aluminum alloy aa5083 using genetic programming. Lecture Notes in Computer Science, pages 421-436, Jan 2021. URL: https://doi.org/10.1007/978-3-030-72699-7\\_27, doi:10.1007/978-3-030-72699-7\\_27.\nPierre Belamri, Henry Proudhon, Damien Texier, and David Ryckelynck. Quaternion-based vision-transformer for polycrystalline ebsd scans pre-trained on large-scale synthetic data. Materials & Design, 258:114599, Oct 2025. URL: https://doi.org/10.1016/j.matdes.2025.114599, doi:10.1016/j.matdes.2025.114599.\nJorge Stella and Michael Pohl. The role of crystal orientation and grain boundaries in the cavitation resistance of en 1.4301 austenitic stainless steel: an ebsd study. Dec 2021. URL: https://doi.org/10.1016/j.wear.2021.204040, doi:10.1016/j.wear.2021.204040.\nChase J. Smith, Xiao-Xiang Yu, Qianying Guo, Christopher R. Weinberger, and Gregory B. Thompson. Phase, hardness, and deformation slip behavior in mixed hf x ta 1-x c. Acta Materialia, 145:142-153, Feb 2018. URL: https://doi.org/10.1016/j.actamat.2017.11.038, doi:10.1016/j.actamat.2017.11.038.\nSeyyed Ezzatollah Moosavi, Cyril Cayron, Yandong Jing, Zeqin Liang, Elisa Cantergiani, Lo\u00efc Aron, Jonathan Friedli, and Roland Log\u00e9. Quantitative correlative sem\u2013ebsd analysis of precipitate\u2013particle and precipitate\u2013boundary interactions in al-mg-si alloys. Materials Characterization, May 2026. URL: https://doi.org/10.1016/j.matchar.2026.116328, doi:10.1016/j.matchar.2026.116328.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_uh7IvP3yQ85eCVKwBxCbjRkg", + "info": { + "start_ts": 619.177332258, + "end_ts": 655.107148383 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. High-strength additively manufacturable Al-Zr-Er-Ni alloys with high as-built ductility and thermal stability\nValid Text Names:\n'ge2025highstrengthadditivelymanufacturable pages 1-2', 'ge2025highstrengthadditivelymanufacturable pages 2-4', 'ge2025highstrengthadditivelymanufacturable pages 4-6', 'ge2025highstrengthadditivelymanufacturable pages 6-7', 'ge2025highstrengthadditivelymanufacturable pages 7-8', 'ge2025highstrengthadditivelymanufacturable pages 8-9', 'ge2025highstrengthadditivelymanufacturable pages 9-10', 'ge2025highstrengthadditivelymanufacturable pages 10-11', 'ge2025highstrengthadditivelymanufacturable pages 11-11'\n\nBibTex:\n@article{ge2025highstrengthadditivelymanufacturable,\n author = \"Ge, Zhaoxuan and Wei, Shaolou and Liu, Zehua and Choma, Tomasz and Zrodowski, Lukasz and Graef, Marc De and Hart, A. John and Taheri-Mousavi, S. Mohadeseh\",\n title = \"High-strength additively manufacturable Al-Zr-Er-Ni alloys with high as-built ductility and thermal stability\",\n year = \"2025\",\n journal = \"Unknown journal\",\n month = \"May\",\n doi = \"10.21203/rs.3.rs-6558867/v1\",\n url = \"https://doi.org/10.21203/rs.3.rs-6558867/v1\",\n publisher = \"Springer Science and Business Media LLC\"\n}\n\n\nAbstract:\nAbstract\n

Additively manufactured (AM) aluminum (Al) alloys with high as-built ductility offer sustainability, enhanced load-bearing performance, and resistance to residual-stress-induced failure. However, most commercial AM Al alloys have very limited ductility. In this study, we designed an AM Al-1.63Zr-2.38Er-1.34Ni (wt. %) alloy with 16% as-built ductility, while not introducing nanopowders or expensive scandium (Sc) to the formulation. We revealed that Al3Zr accounted for 65% of cavitation in the as-built condition by SEM characterizations. With calculation of phase diagram (CALPHAD)-based integrated computational materials engineering (ICME) simulations, we found that at each Er composition, Zr must not exceed the threshold required to act as a reservoir for the L12 phase transformation (needed for high strength) during aging. An optimized composition was identified following the guideline of Zr = 1.07 \u00d7 Er; this composition enhances the ductility from 0.9% to 16% (1600% increase) in the as-built sample, while only decreasing strength by 27% compared to the benchmark alloy. The strength is preserved even after aging for 48 hours at 400\u00b0C.

\n\nRelevant Snippet:\n\u2026 through approaches such as laser parameter optimization21,26\u201333, \u2026 Accurate indexing is important in EBSD quantification \u2026 on indexing fidelity, especially for traditional Hough Indexing \u2026\n\n\n---\n\n## 2. Phase Mapping in EBSD Using Convolutional Neural Networks\nValid Text Names:\n'kaufmann2020phasemappingin pages 1-2', 'kaufmann2020phasemappingin pages 2-3', 'kaufmann2020phasemappingin pages 3-4', 'kaufmann2020phasemappingin pages 4-6', 'kaufmann2020phasemappingin pages 6-7', 'kaufmann2020phasemappingin pages 7-9', 'kaufmann2020phasemappingin pages 9-11', 'kaufmann2020phasemappingin pages 9-9'\n\nBibTex:\n@article{kaufmann2020phasemappingin,\n author = \"Kaufmann, Kevin and Zhu, Chaoyi and Rosengarten, Alexander S. and Maryanovsky, Daniel and Wang, Haoren and Vecchio, Kenneth S.\",\n title = \"Phase Mapping in EBSD Using Convolutional Neural Networks\",\n year = \"2020\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"26\",\n pages = \"458-468\",\n month = \"May\",\n doi = \"10.1017/s1431927620001488\",\n url = \"https://doi.org/10.1017/s1431927620001488\",\n publisher = \"Oxford University Press (OUP)\",\n issue = \"3\",\n issn = \"1431-9276\"\n}\n\n\nAbstract:\nAbstract The emergence of commercial electron backscatter diffraction (EBSD) equipment ushered in an era of information rich maps produced by determining the orientation of user-selected crystal structures. Since then, a technological revolution has occurred in the quality, rate detection, and analysis of these diffractions patterns. The next revolution in EBSD is the ability to directly utilize the information rich diffraction patterns in a high-throughput manner. Aided by machine learning techniques, this new methodology is, as demonstrated herein, capable of accurately separating phases in a material by crystal symmetry, chemistry, and even lattice parameters with fewer human decisions. This work is the first demonstration of such capabilities and addresses many of the major challenges faced in modern EBSD. Diffraction patterns are collected from a variety of samples, and a convolutional neural network, a type of machine learning algorithm, is trained to autonomously recognize the subtle differences in the diffraction patterns and output phase maps of the material. This study offers a path to machine learning coupled phase mapping as databases of EBSD patterns encompass an increasing number of the possible space groups, chemistry changes, and lattice parameter variations.\n\nRelevant Snippet:\n\u2026 , the Hough transform has continued to be the standard in pattern indexing, even though \u2026 The inability for state-of-the-art Hough-based EBSD systems to make determinations about \u2026\n\n\nUnobtainable Papers:\nXiaoxiang Yu, Sivaji Karna, Zachary Haddock, Paul Nolan, and John Carsley. Structure-based phase identification in aluminum alloys using kikuchi band and hough-based ebsd. Materials Today Communications, 50:114476, Jan 2026. URL: https://doi.org/10.1016/j.mtcomm.2025.114476, doi:10.1016/j.mtcomm.2025.114476.\nMiu Lun Lau, Abraham Burleigh, Jeff Terry, and Min Long. Materials characterization: can artificial intelligence be used to address reproducibility challenges? Journal of Vacuum Science & Technology A, Nov 2023. URL: https://doi.org/10.1116/6.0002809, doi:10.1116/6.0002809.\nSrinivas Doddapaneni, Sathish Kumar, Sathyashankara Sharma, Gowri Shankar, Manjunath Shettar, Nitesh Kumar, Ganesha Aroor, and Syed Mansoor Ahmad. Advancements in ebsd techniques: a comprehensive review on characterization of composites and metals, sample preparation, and operational parameters. Journal of Composites Science, 9:132, Mar 2025. URL: https://doi.org/10.3390/jcs9030132, doi:10.3390/jcs9030132.\nYe Wei, Baptiste Gault, Rama Srinivas Varanasi, Dierk Raabe, Michael Herbig, and Andrew J. Breen. Machine-learning-based atom probe crystallographic analysis. Ultramicroscopy, 194:15-24, Nov 2018. URL: https://doi.org/10.1016/j.ultramic.2018.06.017, doi:10.1016/j.ultramic.2018.06.017.\nYuling Yang, Feng Jiang, M. Yang, X. M. Li, H. G. Zheng, and QM Zhang. Electron backscatter diffraction analysis of strain distribution in adiabatic shear band and its nearby area in ti\u20133al\u20135mo\u20134\u00b75v alloy. Materials Science and Technology, 28:165-170, Feb 2012. URL: https://doi.org/10.1179/1743284711y.0000000007, doi:10.1179/1743284711y.0000000007.\nAnalysis and microstructure based simulation of fatigue damage in high-performance steels\nF. Niessen, A. Burrows, and A. Bastos da Silva Fanta. A systematic comparison of on-axis and off-axis transmission kikuchi diffraction. Ultramicroscopy, 186:158-170, Mar 2018. URL: https://doi.org/10.1016/j.ultramic.2017.12.017, doi:10.1016/j.ultramic.2017.12.017.\nKaustubh Venkatraman, Meriem Ben Haj Slama, Vincent Taupin, Nabila Maloufi, and Antoine Guitton. Tuning critical resolved shear stress ratios for bcc-titanium ti21s via an automated data analysis approach. Modelling and Simulation in Materials Science and Engineering, 29:055014, Jun 2021. URL: https://doi.org/10.1088/1361-651x/abfeb0, doi:10.1088/1361-651x/abfeb0.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_PGusrK3loi2SP0YX9X3dJxyb", + "info": { + "start_ts": 619.212901675, + "end_ts": 655.08205647 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Artificial Intelligence and Machine Learning in Cold Spray Additive Manufacturing: A Systematic Literature Review\nValid Text Names:\n'afsharnia2025artificialintelligenceand pages 1-2', 'afsharnia2025artificialintelligenceand pages 7-8', 'afsharnia2025artificialintelligenceand pages 8-9', 'afsharnia2025artificialintelligenceand pages 9-10', 'afsharnia2025artificialintelligenceand pages 10-11', 'afsharnia2025artificialintelligenceand pages 11-12', 'afsharnia2025artificialintelligenceand pages 12-13', 'afsharnia2025artificialintelligenceand pages 14-14', 'afsharnia2025artificialintelligenceand pages 15-16', 'afsharnia2025artificialintelligenceand pages 16-17', 'afsharnia2025artificialintelligenceand pages 17-18', 'afsharnia2025artificialintelligenceand pages 18-20', 'afsharnia2025artificialintelligenceand pages 21-23', 'afsharnia2025artificialintelligenceand pages 26-27', 'afsharnia2025artificialintelligenceand pages 30-31', 'afsharnia2025artificialintelligenceand pages 33-34', 'afsharnia2025artificialintelligenceand pages 34-36', 'afsharnia2025artificialintelligenceand pages 36-38', 'afsharnia2025artificialintelligenceand pages 38-39', 'afsharnia2025artificialintelligenceand pages 40-42', 'afsharnia2025artificialintelligenceand pages 42-44', 'afsharnia2025artificialintelligenceand pages 44-45', 'afsharnia2025artificialintelligenceand pages 45-46', 'afsharnia2025artificialintelligenceand pages 50-52', 'afsharnia2025artificialintelligenceand pages 55-56'\n\nBibTex:\n@article{afsharnia2025artificialintelligenceand,\n author = \"Afsharnia, Habib and Butt, Javaid\",\n title = \"Artificial Intelligence and Machine Learning in Cold Spray Additive Manufacturing: A Systematic Literature Review\",\n year = \"2025\",\n journal = \"Journal of Manufacturing and Materials Processing\",\n volume = \"9\",\n pages = \"334\",\n month = \"Oct\",\n doi = \"10.3390/jmmp9100334\",\n url = \"https://doi.org/10.3390/jmmp9100334\",\n publisher = \"MDPI AG\",\n issue = \"10\",\n issn = \"2504-4494\"\n}\n\n\nAbstract:\nDue to its unique benefits over conventional subtractive manufacturing, additive manufacturing methods continue to attract interest in both academia and industry. One such method is called Cold Spray Additive Manufacturing (CSAM), a solid-state coating deposition technology to manufacture repair metallic components using a gas jet and powder particles. CSAM offers low heat input, stable phases, suitability for heat-sensitive substrates, and high deposition rates. However, persistent challenges include porosity control, geometric accuracy near edges and concavities, anisotropy, and cost sensitivities linked to gas selection and nozzle wear. Interdisciplinary research across manufacturing science, materials characterisation, robotics, control, artificial intelligence (AI), and machine learning (ML) is deployed to overcome these issues. ML supports quality prediction, inverse parameter design, in situ monitoring, and surrogate models that couple process physics with data. To demonstrate the impact of AI and ML on CSAM, this study presents a systematic literature review to identify, evaluate, and analyse published studies in this domain. The most relevant studies in the literature are analysed using keyword co-occurrence and clustering. Four themes were identified: design for CSAM, material analytics, real-time monitoring and defect analytics, and deposition and AI-enabled optimisation. Based on this synthesis, core challenges are identified as small and varied datasets, transfer and identifiability limits, and fragmented sensing. Main opportunities are outlined as physics-based surrogates, active learning, uncertainty-aware inversion, and cloud-edge control for reliable and adaptable ML use in CSAM. By systematically mapping the current landscape, this work provides a critical roadmap for researchers to target the most significant challenges and opportunities in applying AI/ML to industrialise CSAM.\n\nRelevant Snippet:\n\u2026 CSAM parameter selection as optimisation over fast forward \u2026 with a particle swarm optimiser to retrieve impact angle, particle \u2026 Electron backscatter diffraction (EBSD) adds another layer \u2026\n\n\nUnobtainable Papers:\nBin Huang, Hengrui Li, Yugang Miao, Yuyang Zhao, Ji Liu, and Bintao Wu. Curvature and central angle-based morphology measurement method for digital arc additive manufacturing. Measurement, 257:118917, Jan 2026. URL: https://doi.org/10.1016/j.measurement.2025.118917, doi:10.1016/j.measurement.2025.118917.\nTao Shen and Bo Li. An integrated computation framework for predicting mechanical performance of single-phase alloys manufactured using laser powder bed fusion: a case study of cocrfemnni high-entropy alloy. Jun 2024. URL: https://doi.org/10.1016/j.mtcomm.2024.109180, doi:10.1016/j.mtcomm.2024.109180.\nChangrong Guo, Heng Xiong, Tao Ye, Jianfeng Xu, and Long Bai. Ultrasonic defect characterization in coarse-grained materials using shade-a-tfm with subarray-based parameter optimization. IEEE Transactions on Instrumentation and Measurement, 74:1-18, Jan 2025. URL: https://doi.org/10.1109/tim.2025.3562984, doi:10.1109/tim.2025.3562984.\nBingqing Chen, Longchao Zhuo, Yixing Xie, Shuai Huang, Tianyuan Wang, Taiqi Yan, Xiufang Gong, and Yanlin Wang. Comparative study on microstructure, mechanical and high temperature oxidation resistant behaviors of slm in718 superalloy before and after heat treatment. Jul 2024. URL: https://doi.org/10.1016/j.jmrt.2024.06.163, doi:10.1016/j.jmrt.2024.06.163.\nSong Zhang, Y. C. Lin, Yun-Han Ling, Dao-Guang He, Ming-Song Chen, Miao Wan, Gui-Cheng Wu, Ning-Fu Zeng, Hui-Jie Zhang, and Majid Naseri. Modeling flow behaviors and microstructure evolution of ti55511 alloy during the double-stage hot deformation process utilizing machine learning algorithm. Metals and Materials International, Jun 2026. URL: https://doi.org/10.1007/s12540-025-01995-8, doi:10.1007/s12540-025-01995-8.\nKi-Seong Park, Sujung Son, Hyo-Sun Jang, Jiwon Park, Jin-Woong Lee, Shi Woo Lee, Kee-Sun Sohn, Hyoung Seop Kim, and Chang-Seok Oh. An expandable closed-loop framework for ductility optimization in refractory high-entropy alloys. Materials Today Advances, 29:100712, Mar 2026. URL: https://doi.org/10.1016/j.mtadv.2026.100712, doi:10.1016/j.mtadv.2026.100712.\nUday M. Basheer Al-Naib. Artificial intelligence in friction stir welding of ceramic-reinforced metal composites: a review on process optimization and property prediction. The International Journal of Advanced Manufacturing Technology, 141(3-4):1095-1112, Oct 2025. URL: https://doi.org/10.1007/s00170-025-16803-1, doi:10.1007/s00170-025-16803-1.\nEmmanuel Mart\u00ednez-Guerrero, Pedro Lagos-Eulogio, Pedro Miranda-Romagnoli, Roberto Noriega-Papaqui, and Juan Carlos Seck-Tuoh-Mora. Historical elite differential evolution based on particle swarm optimization algorithm for texture optimization with application in particle physics. Applied Sciences, 14:9110, Oct 2024. URL: https://doi.org/10.3390/app14199110, doi:10.3390/app14199110.\nR. Suresh, A. Sureshkumar, V. K. Shanmuganathan, and P. Michael Joseph Stalin. Deep generative learning for microstructure evolution prediction in multimaterial welded interfaces. Welding in the World, Mar 2026. URL: https://doi.org/10.1007/s40194-026-02410-5, doi:10.1007/s40194-026-02410-5.\nHoda Zamani, Mohammad H Nadimi-Shahraki, Shokooh Taghian, and Mahdis Banaie-Dezfouli. Enhancement of bernstain-search differential evolution algorithm to solve constrained engineering problems. Text, Jan 2020. URL: https://doi.org/10.13140/rg.2.2.16902.40004, doi:10.13140/rg.2.2.16902.40004.\nSabitabrata Bhattacharya, Suman Lata Tripathi, and Vikram Kumar Kamboj. Design of tunnel fet architectures for low power application using improved chimp optimizer algorithm. Engineering with Computers, 39:1415-1458, Nov 2023. URL: https://doi.org/10.1007/s00366-021-01530-4, doi:10.1007/s00366-021-01530-4.\nLi Hu, Shu-yong Jiang, Yan-qiu Zhang, Xiao-ming Zhu, and Dong Sun. Texture evolution and inhomogeneous deformation of polycrystalline cu based on crystal plasticity finite element method and particle swarm optimization algorithm. Journal of Central South University, 24:2747-2756, Dec 2017. URL: https://doi.org/10.1007/s11771-017-3688-1, doi:10.1007/s11771-017-3688-1.\nXue-zhao Wang, Ping Zhang, Xiaomin Jiang, Xue Chen, and Youqiang Wang. Influence of solid-solution processes on the microstructure and dynamic impact behavior of cast mg-gd-y-based alloys. Aug 2025. URL: https://doi.org/10.1016/j.vacuum.2025.114280, doi:10.1016/j.vacuum.2025.114280.\nXingrui Li and Ying Guo. Particle swarm optimization based on cubic chaotic mapping and random differential mutation. Algorithms, 19:297, Apr 2026. URL: https://doi.org/10.3390/a19040297, doi:10.3390/a19040297.\nCh. Leela Kumari, Vikram Kumar Kamboj, S. K. Bath, Suman Lata Tripathi, Megha Khatri, and Shivani Sehgal. A boosted chimp optimizer for numerical and engineering design optimization challenges. Engineering with Computers, 39:1-52, Mar 2023. URL: https://doi.org/10.1007/s00366-021-01591-5, doi:10.1007/s00366-021-01591-5.\nChaoyi Zhu, Christian Kurniawan, Marcus Ochsendorf, Dayong An, Stefan Zaefferer, and Marc De Graef. Orientation, pattern center refinement and deformation state extraction through global optimization algorithms. Ultramicroscopy, 233:113407, Mar 2022. URL: https://doi.org/10.1016/j.ultramic.2021.113407, doi:10.1016/j.ultramic.2021.113407.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_KI8Kw21UgTxchIJAACgrtWy6", + "info": { + "start_ts": 619.245570574, + "end_ts": 655.096975494 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nB. S\u00f8rensen, J. Hjelen, H W \u00c5nes, and T. Breivik. Recent features in ebsd, including new trapezoidal correction for multi-mapping. IOP Conference Series: Materials Science and Engineering, 891:012021, Aug 2020. URL: https://doi.org/10.1088/1757-899x/891/1/012021, doi:10.1088/1757-899x/891/1/012021.\nN.S. De Vincentis and D.P. Field. Factors affecting confidence index in ebsd analysis. Ultramicroscopy, 225:113269, Jun 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113269, doi:10.1016/j.ultramic.2021.113269.\nMatthew M Nowell and D Parker Mancuso. Revisiting the effects of sample preparation on ebsd performance in the era of forward model indexing. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.319, doi:10.1093/mam/ozaf048.319.\nYitian Shen, Yongsheng Zhang, Wei Li, Hong Miao, Yongzhe Wang, and Yi Zeng. Local kikuchi band detection in electron backscatter diffraction patterns for enhanced pattern indexing. Nov 2021. URL: https://doi.org/10.1111/jmi.13061, doi:10.1111/jmi.13061.\nJonathan Ligda. Improving polytype identification of silicon carbide using dictionary indexing. ArXiv, Oct 2023. URL: https://doi.org/10.21236/ad1214125, doi:10.21236/ad1214125.\nEdward L. Pang, Peter M. Larsen, and Christopher A. Schuh. Resolving pseudosymmetry in tetragonal zro2 using electron backscatter diffraction with a modified dictionary indexing approach. Journal of Applied Crystallography, 53:1060-1072, Jul 2020. URL: https://doi.org/10.1107/s160057672000864x, doi:10.1107/s160057672000864x.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_AFeXRuxf8pxFGq9DkB0yS3lq", + "info": { + "start_ts": 619.277339244, + "end_ts": 655.062107148 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.\nValid Text Names:\n'burnett2016largevolumeserial pages 1-2', 'burnett2016largevolumeserial pages 2-3', 'burnett2016largevolumeserial pages 3-4', 'burnett2016largevolumeserial pages 4-6', 'burnett2016largevolumeserial pages 6-7', 'burnett2016largevolumeserial pages 7-10', 'burnett2016largevolumeserial pages 10-10', 'burnett2016largevolumeserial pages 10-11', 'burnett2016largevolumeserial pages 11-11'\n\nBibTex:\n@article{burnett2016largevolumeserial,\n author = \"Burnett, Timothy L. and Burnett, Timothy L. and Kelley, R. and Winiarski, B. and Contreras, L. and Daly, Michael and Gholinia, A. and Burke, M. G. and Withers, P. J.\",\n title = \"Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.\",\n year = \"2016\",\n journal = \"Ultramicroscopy\",\n volume = \"161\",\n pages = \"119-129\",\n month = \"Feb\",\n doi = \"10.1016/j.ultramic.2015.11.001\",\n url = \"https://doi.org/10.1016/j.ultramic.2015.11.001\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n\n\nRelevant Snippet:\n\u2026 with a dwell time=6 \u00b5s and a single frame acquisition taking 3 \u2026 equivalent quality as determined by EBSD pattern quality from \u2026 optimisation it may be possible to push the milling currents \u2026\n\n\n---\n\n## 2. Correction of artefacts associated with large area EBSD\nValid Text Names:\n'winiarski2021correctionofartefacts pages 1-4', 'winiarski2021correctionofartefacts pages 4-6', 'winiarski2021correctionofartefacts pages 6-9', 'winiarski2021correctionofartefacts pages 9-11', 'winiarski2021correctionofartefacts pages 11-15', 'winiarski2021correctionofartefacts pages 15-17', 'winiarski2021correctionofartefacts pages 17-19', 'winiarski2021correctionofartefacts pages 19-21', 'winiarski2021correctionofartefacts pages 21-22'\n\nBibTex:\n@misc{winiarski2021correctionofartefacts,\n author = \"Winiarski, B. and Gholinia, A. and Mingard, K. and Gee, M. and Thompson, G. and Withers, P.J.\",\n title = \"Correction of artefacts associated with large area EBSD\",\n year = \"2021\",\n journal = \"Ultramicroscopy\",\n volume = \"226\",\n pages = \"113315\",\n month = \"Jul\",\n doi = \"10.1016/j.ultramic.2021.113315\",\n url = \"https://doi.org/10.1016/j.ultramic.2021.113315\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n\n\nRelevant Snippet:\n\u2026 This means EBSD data are often mapped at a dwell time per \u2026 quality of the EBSD patterns. Subsequently, SE images were \u2026 Fourthly, a verification test is run to optimise the q i spacing \u2026\n\n\nUnobtainable Papers:\nMeiyue Shao, Sriram Vijayan, Peeyush Nandwana, and Joerg R. Jinschek. The effect of beam scan strategies on microstructural variations in ti-6al-4v fabricated by electron beam powder bed fusion. Nov 2020. URL: https://doi.org/10.1016/j.matdes.2020.109165, doi:10.1016/j.matdes.2020.109165.\nGert Nolze, Maria J\u00fcrgens, J\u00fcrgen Olbricht, and Aimo Winkelmann. Improving the precision of orientation measurements from technical materials via ebsd pattern matching. Acta Materialia, 159:408-415, Oct 2018. URL: https://doi.org/10.1016/j.actamat.2018.08.028, doi:10.1016/j.actamat.2018.08.028.\n\u0160t\u011bp\u00e1n Gamanov, Poczkl\u00e1n Ladislav, Antonin Dlouhy, and Tomas Kruml. A step size dependence of kam, grod and gnd parameters calculated from ebsd. Jan 2025. URL: https://doi.org/10.2139/ssrn.4789333, doi:10.2139/ssrn.4789333.\nRobert A. Schwarzer and Jarle Hjelen. High-speed orientation microscopy with offline solving sequences of ebsd patterns. Solid State Phenomena, 160:295-300, Feb 2010. URL: https://doi.org/10.4028/www.scientific.net/ssp.160.295, doi:10.4028/www.scientific.net/ssp.160.295.\nYongjun Chen, J. Hjelen, S. Gireesh, and H. Roven. Optimization of ebsd parameters for ultra\u2010fast characterization. Journal of Microscopy, 245:111-118, Feb 2012. URL: https://doi.org/10.1111/j.1365-2818.2011.03551.x, doi:10.1111/j.1365-2818.2011.03551.x.\nStuart I. Wright, Matthew M. Nowell, Ren\u00e9 de Kloe, Patrick Camus, and Travis Rampton. Electron imaging with an ebsd detector. Ultramicroscopy, 148:132-145, Jan 2015. URL: https://doi.org/10.1016/j.ultramic.2014.10.002, doi:10.1016/j.ultramic.2014.10.002.\nPascal Thome, Micheal Madigan, Shadman Tahsin Nabil, Owen Lowerey, Francisco Medina, and Sammy Tin. Columnar to equiaxed transition in electron beam powder bed fusion (eb\u2010pbf) characterized through microstructure analytics. Advanced Engineering Materials, Mar 2026. URL: https://doi.org/10.1002/adem.202503160, doi:10.1002/adem.202503160.\nM. M. Nowell and S. I. Wright. Phase differentiation via combined ebsd and xeds. Journal of Microscopy, 213:296-305, Mar 2004. URL: https://doi.org/10.1111/j.0022-2720.2004.01299.x, doi:10.1111/j.0022-2720.2004.01299.x.\nT. Arold, P. Kroo\u00df, and T. Niendorf. Pathways to exploit the multi-spot scanning strategy in electron beam additive manufacturing for control of microstructure and defect density. Jul 2024. URL: https://doi.org/10.1016/j.jmrt.2024.06.043, doi:10.1016/j.jmrt.2024.06.043.\nE. Uta, N. Gey, P. Bocher, M. Humbert, and J. Gilgert. Texture heterogeneities in \u03b1p/\u03b1s titanium forging analysed by ebsd\u2010relation to fatigue crack propagation. Journal of Microscopy, 233:451-459, Mar 2009. URL: https://doi.org/10.1111/j.1365-2818.2009.03141.x, doi:10.1111/j.1365-2818.2009.03141.x.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_rjeQSC8BesdXWTotR8LKnLxa", + "info": { + "start_ts": 619.325013734, + "end_ts": 655.072304035 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\nValid Text Names:\n'stanek2020comparisonofsegmentation pages 1-2', 'stanek2020comparisonofsegmentation pages 2-7', 'stanek2020comparisonofsegmentation pages 7-9', 'stanek2020comparisonofsegmentation pages 9-12', 'stanek2020comparisonofsegmentation pages 12-16', 'stanek2020comparisonofsegmentation pages 16-18', 'stanek2020comparisonofsegmentation pages 18-19'\n\nBibTex:\n@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n\n\nAbstract:\nA comparison study of grain segmentation in 2D and 3D electron backscatter diffraction (EBSD) measurements of polycrystalline materials, based on a misorientation threshold, is presented. Using Laguerre tessellations with orientation marks, a simulation study that demonstrates a different sensitivity of segmentation in the case of low angle and high angle grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investigated concerning segmentation, and there are given some recommendations for the misorientation thresholds. Computational tools such as DREAM 3D software enable significant data evaluation that is completed by a five-parameter analysis of grain boundaries. K e y w o r d s: electron backscatter diffraction (EBSD), grain boundaries, misorientation, tessellation\n\nRelevant Snippet:\n\u2026 Computational tools such as DREAM 3D software enable \u2026 of grains, in a single 2D slice, on the misorientation tolerance level\u2026 number of grains in a slice with the misorientation threshold \u2026\n\n\n---\n\n## 2. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\nValid Text Names:\n'folwarczny2026accurategrainboundary pages 1-3', 'folwarczny2026accurategrainboundary pages 3-5', 'folwarczny2026accurategrainboundary pages 5-9', 'folwarczny2026accurategrainboundary pages 9-12', 'folwarczny2026accurategrainboundary pages 12-15', 'folwarczny2026accurategrainboundary pages 15-18', 'folwarczny2026accurategrainboundary pages 18-20'\n\nBibTex:\n@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n\n\nRelevant Snippet:\n\u2026 Two grain boundaries displaying the same misorientation between adjacent \u2026 normally encountered in experimental EBSD data. From the grain boundary segments, GBPDs were \u2026\n\n\nUnobtainable Papers:\nWilliam C. Lenthe, McLean P. Echlin, Jean C. Stinville, Marc De Graef, and Tresa M. Pollock. Twin related domain networks in ren\u00e9 88dt. Materials Characterization, 165:110365, Jul 2020. URL: https://doi.org/10.1016/j.matchar.2020.110365, doi:10.1016/j.matchar.2020.110365.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_x3Q5do7NWtrRZiM5mvwykX6y", + "info": { + "start_ts": 619.362055206, + "end_ts": 655.052245472 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_GpTgn8r9BIdram9mIMRTK5Lk", + "type": "function", + "function": { + "arguments": "{\"question\": \"Identify explicit evidence relevant to any of these: (1) automated optimizer-driven tuning of EBSD pattern processing/indexing parameters; (2) iterative re-indexing of stored raw patterns with CI/IQ/fit/MAD/NCC objective selecting next parameters; (3) parameter sweeps or optimization using grain-level outcomes; (4) hybrid Hough then dictionary/pattern matching on low-quality residue with automated escalation criterion; (5) autonomous EBSD post-processing or feedback to SEM acquisition parameters. For each work report exact method, automated variables/objective, and limitations/non-automated parts.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "No relevant evidence found for your question.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=24 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_GpTgn8r9BIdram9mIMRTK5Lk", + "info": { + "start_ts": 668.979169114, + "end_ts": 675.651561975 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_AqiInZ3vFforOEM02yK3ZwoA", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"Reindexing\\\" EBSD \\\"evolutionary algorithm\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_RgOFyPWBn5U8Hvx3JRcTYJut", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"dictionary indexing\\\" EBSD \\\"evolutionary algorithm\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_V0DFMvNN6qIFHJFPJYNQ3VAm", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"global optimization algorithms\\\" EBSD orientation pattern center refinement deformation state extraction\"}", + "name": "paper_search" + } + }, + { + "id": "call_ukRTrcFmrr3QCZ78ACrh54nw", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"EBSD pattern center\\\" optimization normalized cross correlation\"}", + "name": "paper_search" + } + }, + { + "id": "call_viEbTlkbaVWSj9viiSsVoQeq", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"low confidence\\\" EBSD \\\"dictionary indexing\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_sNECA4HSwXwytqITdKKFiRzx", + "type": "function", + "function": { + "arguments": "{\"query\": \"EBSD automated parameter sweep indexing confidence index Hough settings\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_AqiInZ3vFforOEM02yK3ZwoA", + "info": { + "start_ts": 689.614977119, + "end_ts": 703.90877586 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nToby Francis, Paul F. Rottmann, Andrew T. Polonsky, Marie-Agathe Charpagne, McLean P. Echlin, Veronica Anghel, David R. Jones, George T. Gray, Marc De Graef, and Tresa M. Pollock. Multimodal 3d characterization of voids in shock-loaded tantalum: implications for ductile spallation mechanisms. Acta Materialia, 215:117057, Aug 2021. URL: https://doi.org/10.1016/j.actamat.2021.117057, doi:10.1016/j.actamat.2021.117057.\nJ.C. Stinville, M.A. Charpagne, R. Maa\u00df, H. Proudhon, W. Ludwig, P.G. Callahan, F. Wang, I.J. Beyerlein, M.P. Echlin, and T.M. Pollock. Insights into plastic localization by crystallographic slip from emerging experimental and numerical approaches. Annual Review of Materials Research, 53:275-317, Jul 2023. URL: https://doi.org/10.1146/annurev-matsci-080921-102621, doi:10.1146/annurev-matsci-080921-102621.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_RgOFyPWBn5U8Hvx3JRcTYJut", + "info": { + "start_ts": 689.651392441, + "end_ts": 719.573874281 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_V0DFMvNN6qIFHJFPJYNQ3VAm", + "info": { + "start_ts": 689.685870397, + "end_ts": 710.360182466 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nDavid Fullwood and Brent Adams. High resolution orientation imaging microscopy. ArXiv, May 2012. URL: https://doi.org/10.21236/ada586789, doi:10.21236/ada586789.\nSimon Breumier. Methodology for shot-peening induced intragranular residual stress prediction. PhD thesis, Agence Bibliographique de l\u2019Enseignement Sup\u00e9rieur, 2021. URL: https://doi.org/10.70675/9bdea3e2z19e1z4843zae96z7086104f4cb7, doi:10.70675/9bdea3e2z19e1z4843zae96z7086104f4cb7.\nMatthew J. Burch, Chris M. Fancher, Srikanth Patala, Marc De Graef, and Elizabeth C. Dickey. Mapping 180\u00b0 polar domains using electron backscatter diffraction and dynamical scattering simulations. Ultramicroscopy, 173:47-51, Feb 2017. URL: https://doi.org/10.1016/j.ultramic.2016.11.013, doi:10.1016/j.ultramic.2016.11.013.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_ukRTrcFmrr3QCZ78ACrh54nw", + "info": { + "start_ts": 689.717413937, + "end_ts": 713.509185713 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\nValid Text Names:\n'shen2019convolutionalneuralnetworkbased pages 1-2', 'shen2019convolutionalneuralnetworkbased pages 2-3', 'shen2019convolutionalneuralnetworkbased pages 3-5', 'shen2019convolutionalneuralnetworkbased pages 5-6', 'shen2019convolutionalneuralnetworkbased pages 6-8', 'shen2019convolutionalneuralnetworkbased pages 8-10', 'shen2019convolutionalneuralnetworkbased pages 10-12', 'shen2019convolutionalneuralnetworkbased pages 12-13', 'shen2019convolutionalneuralnetworkbased pages 13-14', 'shen2019convolutionalneuralnetworkbased pages 14-14'\n\nBibTex:\n@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n\n\nRelevant Snippet:\n\u2026 -art dictionary indexing (DI) method significantly improves the orientation reconstruction accuracy from highly noisy EBSD \u2026 For those with low confidence, run the full dictionary search. \u2026\n\n\nUnobtainable Papers:\nDavid P. Field, Satyam Suwas, and Stuart I. Wright. Basics of SEM-Based Techniques for Electron Backscatter Diffraction, pages 27-41. Springer Nature Singapore, Jan 2025. URL: https://doi.org/10.1007/978-981-96-5346-1\\_2, doi:10.1007/978-981-96-5346-1\\_2.\nFarangis Ram and Marc De Graef. Phase differentiation by electron backscatter diffraction using the dictionary indexing approach. Acta Materialia, 144:352-364, Feb 2018. URL: https://doi.org/10.1016/j.actamat.2017.10.069, doi:10.1016/j.actamat.2017.10.069.\nAhmed A. Tiamiyu, Tyler Lucas, Edward L. Pang, Xi Chen, James M. LeBeau, and Christopher A. Schuh. Heterogeneous microstructural evolution during hydrodynamic penetration of a high-velocity copper microparticle impacting copper. Jan 2024. URL: https://doi.org/10.1016/j.mattod.2023.11.015, doi:10.1016/j.mattod.2023.11.015.\nDevi Janani Ramesh, Janis S\u00e4lker, Sameer Aman Salman, Eduardo Vega Halleslevens, Moritz to Baben, Christina Scheu, and Jochen M. Schneider. 2d vs. 3d phase analysis of an oxidized cr2alc coating. Unknown journal, Jan 2026. URL: https://doi.org/10.2139/ssrn.6989225, doi:10.2139/ssrn.6989225.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_viEbTlkbaVWSj9viiSsVoQeq", + "info": { + "start_ts": 689.749615248, + "end_ts": 718.236611284 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nSaransh Singh. Application of forward modeling to materials characterization. Text, Jan 2018. URL: https://doi.org/10.1184/r1/6714788.v1, doi:10.1184/r1/6714788.v1.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0497 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_sNECA4HSwXwytqITdKKFiRzx", + "info": { + "start_ts": 689.780833706, + "end_ts": 719.174439903 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_gQXQyhWpcdQAtqZMTz266Aco", + "type": "function", + "function": { + "arguments": "{\"question\": \"From the retrieved papers, extract exact evidence on: Shen et al. 2019 low-confidence patterns and full dictionary search; Polonsky et al. 2019 BisQue/DREAM.3D parameter sweeps and grain segmentation outcomes; global/evolutionary optimization in EBSD pattern matching or pattern-center/orientation refinement; and whether any method iteratively tunes Hough/background/binning/neighbor settings or closes feedback to SEM acquisition. State what is and is not automated.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "No relevant evidence found for your question.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=25 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_gQXQyhWpcdQAtqZMTz266Aco", + "info": { + "start_ts": 732.745937759, + "end_ts": 740.353156312 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_4klcSpImEsXAZeSqWkaHiygh", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"shen2019convolutionalneuralnetworkbased pages 1-2\", \"shen2019convolutionalneuralnetworkbased pages 8-10\", \"shen2019convolutionalneuralnetworkbased pages 10-12\", \"polonsky2019threedimensionalanalysisand pages 1-2\", \"polonsky2019threedimensionalanalysisand pages 5-6\", \"polonsky2019threedimensionalanalysisand pages 6-8\"]}", + "name": "read_text" + } + }, + { + "id": "call_G0lw5LzH1ACLfx203yBWOksS", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"stinville2022multimodaldatasetof pages 9-12\", \"stinville2022multimodaldatasetof pages 12-13\", \"sparks20243dreconstructionof pages 9-12\", \"sparks20243dreconstructionof pages 23-24\", \"kaufmann2020phasemappingin pages 1-2\", \"kaufmann2020phasemappingin pages 6-7\"]}", + "name": "read_text" + } + }, + { + "id": "call_mH0dRIstCN2AOJxkl8ecCACI", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"burnett2016largevolumeserial pages 4-6\", \"burnett2016largevolumeserial pages 6-7\", \"winiarski2021correctionofartefacts pages 4-6\", \"winiarski2021correctionofartefacts pages 15-17\", \"stanek2020comparisonofsegmentation pages 1-2\", \"stanek2020comparisonofsegmentation pages 7-9\"]}", + "name": "read_text" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "## 1. polonsky2019threedimensionalanalysisand pages 6-8 (Context ID: pqac-00000012)\n\n the\n Keeping all other pipeline parameters constant (mini- optimal number of clusters in a dataset. The derivatives are\nmum number of two neighbors, minimum allowed defect approximated via the central difference method, shown in\nsize of 125 voxels), the misorientation tolerance was tested Fig. 5b and c. A best balance between these metrics occurs\nover a wide range of values, from 0.1 to 10\u25e6of mis- at 2.0\u25e6, making this an equitable choice of misorientation\norientation in 0.1\u25e6increments, generating a total of one- tolerance for segmentation of this dataset to avoid excess\nhundred 3D reconstructions of the steel sample described consolidation of separable grains. This results in a dataset\nin \u201cLENS-Processed 304L Stainless Steel.\u201d This represents with 899 grains, wherein 98.4% of the masked data is\nsome seventy gigabytes of generated data. To assess the included in the reconstruction.\nquality of the reconstruction, two high-level metrics were\nused, the fraction of \u201cgood\u201d data, and the total number Subvolume Creation and Analysis of Large 3D\nof grains found during the reconstruction. \u201cGood\u201d data is Volumes\ndefined by the confidence index of the EBSD data, and\ntherefore contains a constant number of voxels that does not Beyond investigation of individual pipeline parameters,\nchange between reconstructions. During the reconstruction the BisQue framework also enables coupled parameteriza-\nprocess, various cleanup steps are employed which retain tion, wherein sets of variables are linked to each other.\nvarying amounts of this initial group of voxels. DREAM.3D This allows for automated generation of random samples\nfilters such as \u201cMinimum Size,\u201d \u201cMinimum Number of of equal-sized subvolumes of 3D datasets, among otherIntegr Mater Manuf Innov (2019) 8:37\u201351 43\n\n\nFig. 4 a A single feature\nidentified from the bulk using a (a)\nmisorientation tolerance of 5.0\u25e6 222 m\nfor segmentation via CCA.\n 420 mGrain orientation is shown in\nIPF coloring with the build\ndirection taken as the reference\ndirection. b The same feature is\nsplit into three separate grains\nwhen a misorientation tolerance\nof 2.0\u25e6is used for segmentation\n\n\n\n\n\n Build\n Direction\n\n\n\n\n 111\n\n\n\n\n\n 001 101\n\n (b)\n\n\n\n\n\n Grain ID\n\n\n\n 1 2 3\n\n\napplications. The characterization of subvolumes in 3D Morphological Complexity of As-printed Additive\ndatasets has been previously applied to define microstruc- Structures\ntural and property volume elements (MVEs and PVEs) for\nvarious material properties, to inform component design Due to the complex thermal gradients and remelting\n[41, 42]. phenomena inherent to additive manufacturing methods,44 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n (a) in standard analyses of 2D cross-sections [47]. Columnar\n grains near the center of melt pool, when cut perpendicu-\n larly, can appear as fine equiaxed grains, which may mimic\n traditional weld microstructures and can lead to misinter-\n pretation of data. Misinterpretation is also highly likely\n for larger grains in additive processes. Either as a result\n of multiple remelting and solidification cycles, or from a\n single solidification event with complex thermal gradients,\n individual grains are commonly observed to possess large\n misorientation gradients. One such large grain was identi-\n fied in the reconstruction of the 304L stainless steel sample\n described in \u201cMisorientation Tolerance Parameterization\n for Feature Identification.\u201d Even while using the previously\n determined misorientation tolerance of 2\u25e6for grain segmen-\n tation, large changes in orientation are observed, in excess\n of 12\u25e6from the average orientation of the grain. Changes in\n orientation are evident via differences in inverse pole figure\n (IPF) coloring of the grain shown in Fig. 6.\n The limitations of 2D characterization of additive\n structures such as those shown in Fig. 6 are demonstrated\n by examining 2D cross-sections from the 3D data volume.\n Rate of Change per Degree\n Using the coupled parameterization ability of BisQue and\n (b)\n the \u201cCrop Geometry\u201d filter in DREAM.3D, individual\n subvolumes, each 1 voxel thick (to approximate 2D cross-\n sections), were extracted from a 3D grain along the three\n principal sample directions (Fig. 6b\u2013d). Figure 6d shows\n Z-normal slices, which are those taken with a plane\n normal along the build direction of the sample, which\n is also parallel to the TriBeam sectioning direction. X-\n normal (Fig. 6b) and Y-normal (Fig. 6c) slices are those\n containing a plane normal along either of the in-plane axes\n as defined by the TriBeam sectioning direction and the\n (c) sample geometry. These slices were then segmented using\n the same misorientation tolerance used on the full volume,\n 2\u25e6, to determine how the lack of 3D data would affect\n the characterization of such a grain. As this analysis was\n only performed on voxels associated with a large feature\n from the 3D volume, features resulting from this additional\n segmentation are subfeatures of the initial 3D grain. An\n example of a Z-normal slice characterization is shown in\n Fig. 7. Since small features comprised of fewer than 64\n voxels were already removed from the 3D volume, a modest\n minimum size requirement of four pixels was applied to\nFig. 5 a The fraction of the voxels in the confidence index mask\n these slices, effectively removing what would normally beincluded in the segmentation and the total number of grains, both as\na function of the defined grain segmentation misorientation tolerance. treated as noisy pixels in typical 2D cross-sections. The 3D\nThe rate of change per degree of misorientation tolerance for b the \n\n## 2. polonsky2019threedimensionalanalysisand pages 5-6 (Context ID: pqac-00000013)\n\nnd grains containing less than 64 voxels voxels. As with the 304L sample, a minimum CI of 0.1\nwere considered too small to be resolved and were removed. was chosen for the initial mask. Grains were segmented\nThis corresponds to 4 \u00d7 4 \u00d7 4 points across a feature and an using a misorientation tolerance of 2\u25e6using the same\nequivalent diameter of 14.9 \u03bcm, roughly five times smaller segmentation algorithm as explained in \u201cMisorientation\nthan the average grain size determined from 2D analysis Tolerance Parameterization for Feature Identification.\u201d A\ndone prior to 3D characterization. Gaps remaining from minimum size requirement of 125 voxels was applied to\nthese removed features were filled via iterative dilation of delete grains that were too small to be properly charac-\nthe surrounding grains. Indexed EBSD data was uploaded terized. This corresponds to 5\u00d75\u00d75 points across a feature\nto BisQue for analysis after data collection. and an equivalent diameter of 9.3 \u03bcm, roughly five times\n smaller than the average grain size determined from 2D\nEBM Inconel 718 analysis done prior to 3D characterization. Gaps remaining\n from these removed features were filled via iterative dilation\nElectron beam melting (EBM) is a powder-bed additive of the surrounding grains. Indexed EBSD data was uploaded\ntechnique which utilizes an electron beam as a heat source. to BisQue for analysis after data collection.\nA fine layer of powder on the order of 50\u2013100 \u03bcm is\nspread over a build plate, and the electron beam selectively\nmelts the powder. The build plate is then lowered and Results\nthe process is repeated to create complex geometries in\n3D. Unlike laser melting techniques, electron beam melting Misorientation Tolerance Parameterization\nrequires vacuum processing environment, and the electron for Feature Identi\ufb01cation\nbeam heat source can also be used to achieve preheat\ntemperatures in excess of 1000 \u25e6C. A sample of EBM In order to analyze 3D tomography data, individual fea-\nInconel 718 was fabricated with 50-\u03bcm-thick build layers tures, or grains, must be segmented from the reconstructed\nand was characterized in the TriBeam system with 1.5-\u03bcm- volume. When working with 3D EBSD data, orientation\nthick slices as described previously [37], and EBSD data information exists at every voxel in the volume. Straight-\nwas collected to characterize grain texture and morphology. forward segmentation algorithms such as connected compo-\nRoughly 25 GB of data was collected from a total of 204 nent analysis (CCA) are used to identify grains from the 3D\nslices over the course of three days, resulting in a volume data using a misorientation tolerance between adjacent vox-\nof 952.5 \u00d7 501.0 \u00d7 306.0 \u03bcm. Raw EBSD patterns were els. The misorientation tolerance is one of the most critical\nnot collected, but the Hough transforms of the patterns parameters for the segmentation and analysis of EBSD data.\n\n\nFig. 3 Mask of the 3D data from\na sample of LENS-processed 816 m\n304L stainless steel generated by 834 m\nthresholding on the Confidence\nIndex (CI) of EBSD patterns\n\n\n\n\n\n 240 m\n\n\n Build\n Direction\n\n\n\n\n\n Confidence Index\n\n\n\n\n 0.1 1.042 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\nA mask is defined that distinguishes good data from noise in Neighbors,\u201d and \u201cFill Bad Data\u201d can affect whether individ-\nthe dataset by using a threshold on the electron backscatter ual voxels are retained or released from segmented features.\npattern confidence index (CI) [36]. To perform grain seg- At the end of a reconstruction, post-processing scripts are\nmentation, a seed is selected from voxel 0 within the mask used to count the number of voxels belonging to segmented\nvolume and its orientation is compared to each of its nearest features, which will always be a subset of the initial group\nneighbors that are still within the mask (up to six neigh- of \u201cgood\u201d voxels.\nbors in a rectilinear voxelized grid). If the misorientation As can be seen in Fig. 5a, these two metrics tend to have\nbetween two adjacent voxels is below the user-defined mis- opposite trends with increasing misorientation tolerance. As\norientation tolerance, then the voxel is added to the current larger misorientation tolerances are used to segment the\ngrain. This process is repeated until no additional neighbor- data, not only are more grains merged together, decreasing\ning voxels can be added to the grain. A new seed is then the overall number of grains found, but noisier data on\nselected at the smallest voxel index not previously grouped the sample surfaces can also be connected to existing\nand the process continues until all voxels in the mask have grains during segmentation. This trend is true except for\nbeen assigned to some grain ID. the lowest misorientation tolerances, below 0.9\u25e6. At these\n It is essential that an optimal misorientation tolerance lower misorientation tolerances, the minimum grain size\nvalue is used during dataset reconstruction due to the requirement of 64 voxels eliminates many of the grains\nimpact of the misorientation tolerance on the definition found during the segmentation step, and for tolerances\nof grain structure. Two-dimensional analysis of EBSD below 0.4\u25e6, no continuous grains larger than 64 voxels\ndata typically uses a misorientation tolerance of 5\u25e6, which are found, so grain-level metrics do not exist in these\nis generally sufficient for well-annealed or recrystallized reconstructions.\nmicrostructures of various morphologies [38]. However, in An optimized reconstruction will maintain a high fraction\ndeformed or highly textured materials, as are commonly of the masked voxels with high confidence index, while\nobserved in as-built additive structures [39, 40], such also reducing unnecessary merging of grains. The merging\nlarge misorientation tolerances can inadvertently merge of grains when a large misorientation tolerance is selected\ngrains that should otherwise be segmented separately. can be seen in Fig. 4. The feature in Fig. 4a is identified\nThe additional voxel connectivity provided by 3D data as a single grain when using a misorientation tolerance\nfurther exacerbates this problem, and as there is generally of 5.0\u25e6, but becomes separated into three separate grains\nnot an accepted misorientation tolerance that applies to when using a misorientation tolerance of 2.0\u25e6(Fig. 4a).\nall 3D microstructures, these values must be chosen The further separation of this feature results in grains with\nintuitively and the resulting segmentation analyzed for more reasonable grain morphologies that reflect the additive\nefficacy. This is a time-consuming, laborious process, and process. The best misorientation tolerance value is the\nthe ultimate selection of a user-defined parameter is not lowest value at which both the change in fraction of masked\nnecessarily verifiable as the optimal choice. The use of data included and the change in the number of grains\nparameterization, enabled by the BisQue framework, allows became constant, or when the derivatives of these quantities\nfor the generation of many 3D volumes to better assess these with respect to misorientation tolerance begin to converge,\nand other important reconstruction parameters. similar to the Elbow method used for determining the\n Keeping all other pipeline parameters constant (mini- optimal number of clusters in a dataset. The derivatives are\nmum number of two neighbors, minimum allowed defect approximated via the central difference method, shown in\nsize of 125 voxels), the misorientation tolerance was tested Fig. 5b and c. A best balance between these metrics occurs\nover a wide range of values, from 0.1 to 10\u25e6of mis- at 2.0\u25e6, making this an equitable choice of misorientation\norientation in 0.1\u25e6increments, generating a total of one- tolerance for segmentation of this dataset to avoid excess\nhundred 3D reconstructions of the steel sample described consolidation of separable grains. This results in a dataset\nin \u201cLENS-Processed 304L Stainless \n\n## 3. polonsky2019threedimensionalanalysisand pages 1-2 (Context ID: pqac-00000014)\n\nIntegrating Materials and Manufacturing Innovation (2019) 8:37\u201351\nhttps://doi.org/10.1007/s40192-019-00126-7\n\n TECHNICAL ARTICLE\n\n\nThree-dimensional Analysis and Reconstruction of Additively\nManufactured Materials in the Cloud-Based BisQue Infrastructure\n\nAndrew T. Polonsky1 \u00b7 Christian A. Lang2 \u00b7 Kristian G. Kvilekval2 \u00b7 Marat I. Latypov1,3 \u00b7 McLean P. Echlin1 \u00b7\nB. S. Manjunath2 \u00b7 Tresa M. Pollock1\n\n\nReceived: 30 December 2018 / Accepted: 18 February 2019 / Published online: 22 March 2019\n\u00a9 The Minerals, Metals & Materials Society 2019\n\nAbstract\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a\ncloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline\nvariables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-\ncollected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore,\na comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using\nBisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects\nfor impact in material science are also discussed.\n\nKeywords Tomography \u00b7 3D materials science \u00b7 Microstructure analytics \u00b7 Cloud infrastructure \u00b7 Additive manufacturing \u00b7\n3D metal printing\n\n\n Introduction\n Andrew T. Polonsky\n polonsky@ucsb.edu\n Recent external pressures mandated by government grants\n Christian A. Lang requiring data be accessible to the public, and internal\n clang@ucsb.edu pressures within academic communities for data sharing\n and reproducibility have led to the emergence of data-\n Kristian G. Kvilekval\n driven platforms within the scientific community [1\u20134].\n kgk@ucsb.edu\n Requirements for these platforms include providing data\n Marat I. Latypov provenance, data sharing, and hosting services for data that\n latmarat@ucsb.edu may otherwise disappear when a lab closes or hardware\n fails [5]. Recently, data repositories have shifted from\n McLean P. Echlin\n mechlin@ucsb.edu simply siloing data to include data analytics and predictive\n capabilities. Furthermore, some of these repositories are\n B. S. Manjunath directly linked with high-performance computing in order\n manj@ucsb.edu\n to populate an ever growing database, such as with\n Tresa M. Pollock thermodynamic databases such as the Materials Project [6]\n tresap@ucsb.edu and the Open Quantum Materials Database (OQMD) [7].\n More sophisticated predictive/interactive databases, such as\n1 the aforementioned thermodynamic ones, have been slower Materials Department, University of California Santa Barbara,\n Santa Barbara, CA 93106, USA to develop with the 3D Materials Science community. This\n2 is likely due to the less standardized and more diverse data Electrical and Computer Engineering Department,\n University of California Santa Barbara, Santa Barbara, modalities, the complexity and diversity of microstructures,\n CA 93106, USA and the relative infancy of the community itself.\n3 Mechanical Engineering Department, University of California In almost all regards, using 3D data to answer scientific\n Santa Barbara, Santa Barbara, CA 93106, USA questions in Materials Science is challenging due to the38 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\nneed for dataset acquisition, reconstruction, and analysis. In 3D characterization techniques, such as X-ray-based\nReconstruction of 3D datasets is a complex process with diffraction contrast tomography (DCT), the problems of\nmany degrees of freedom. The algorithms employed, and data reconstruction and analysis are also convoluted. Coher-\ntolerances selected by the user to align slices, segment ent volumes must be reconstructed from individual 2D\nfeatures, and remove data artifacts (i.e., remove noise), images via back projection techniques [24]. The diffraction\nultimately determine the quantitative data extracted from spots from the 2D projections are generally assumed to orig-\nthe volume. Significant progress on the reconstruction and inate from a single grain, limiting the characterization of\nanalysis of electron backscatter diffraction (EBSD) data has materials with large internal strain gradients [25]. Moreover,\nbeen made by a software package DREAM.3D [8], which stereological conventions established in 2D do not neces-\ncreates a unified data structure for 3D data treatment and sarily capture the real 3D microstructures, particularly at\naggregates tools and algorithms collectively developed by extrema of property distributions [26, 27]. Given the wealth\nthe community. In DREAM.3D, this process takes the form of additional data provided by 3D characterization, new\nof a pipeline, which typically consists of well over a dozen tools aimed at exploring the parameter space offered to 3D\nsteps, and in which each step has its own set of input reconstruction are required. In the following manuscript, we\nvariables and outputs. DREAM.3D typically runs in a serial describe the use of a cloud-based platform, BisQue [28, 29],\nprocess on a workstation, with predefined input parameters. enhanced with the DREAM.3D microstructural software\nDue to the long computation times required to process analysis package for the 3D reconstruction and analysis\n3D data, which often scales proportionately as dataset size of additively manufactured microstructures characterized\nand complexity increases, it is impractical to explore the using TriBeam tomography [30].\nfull parameter space by manually running reconstructions.\nTypically, only several variables are adjusted by a user until\na qualitatively acceptable reconstruction is attained. This Methods\nprocess inherently imparts bias and typically provides little\nopportunity to justify specific reconstruction variables. The following sections describe the software infrastructure\n The bias generated during reconstruction of 3D datasets and materials investigated.\nis subsequently carried through to data analysis, which,\nfor microstructural data, is still an ill-defined process. BisQue for Cloud-Based Data Storage and Analysis\nStandardization of analysis involving new characterization\ntechniques is generally a slow process. Automated orien- All data processing was performed in the cloud with BisQue\ntation imaging analysis via EBSD first emerged in the [28, 29] using a web browser interface to interact with the\nearly 90s, but standards did not emerge for nearly two 3D datasets. BisQue has backend infrastructure support for\ndecades [9, 10]. These new standards covered basic quan- web services such as Amazon S3, IRODS, and federated\ntitative analysis, including methods for grain size measure- local file systems as well as client facing tools such as\nment, as well as guidelines governing data quality, quan- Java, Python, and Matlab. A DREAM.3D [8] plugin was\ntity, and resolution sufficient to make such measurements developed for BisQue in order to incorporate the materials\n[11]. Even in 2D EBSD analysis, such standards are rare science based 3D dataset analysis tools. The database\nand do not necessarily address additional subtleties arising architecture of BisQue is such that every processing\nfrom more complex, albeit commonly studied, microstruc- execution, dataset (input and output), and workflow\ntures, including twinned materials, nanocrystalline mate- processing pipeline receives a unique web address URL.\nrials, and deformed materials [12\u201317]. Standardization of These URLs are useful for at least two reasons: tracking\ncharacterization meth\n\n## 4. shen2019convolutionalneuralnetworkbased pages 1-2 (Context ID: pqac-00000015)\n\n Acta Materialia 170 (2019) 118e131\n\n\n Contents lists available at ScienceDirect\n\n Acta Materialia\n\n\n journal homepage: www.elsevier.com/locate/actamat\n\n\nFull length article\n\nConvolutional neural network-based method for real-time orientation\nindexing of measured electron backscatter diffraction patterns\n\nYu-Feng Shen a, b, Reeju Pokharel a, *, Thomas J. Nizolek a, Anil Kumar a, Turab Lookman a\n\na Los Alamos National Laboratory, Los Alamos, NM, 87544, USA\nb Carnegie Mellon University, Pittsburgh, PA, 15213, USA\n\n\n\na r t i c l e i n f o a b s t r a c t\n\n\nArticle history: Electron backscatter diffraction (EBSD) is the most commonly used technique for obtaining spatially\nReceived 13 September 2018 resolved microstructural information from polycrystalline materials. We have developed two convolu-\nReceived in revised form tional neural network approaches based on domain transform and transfer learning to reconstruct\n13 March 2019\n crystal orientations from electron backscatter diffraction patterns. Our models are robust to experi-\nAccepted 20 March 2019\n mentally measured image noise and index orientations as fast as the highest EBSD scanning rates. We\nAvailable online 27 March 2019\n demonstrate that the quaternion norm metric is a strong indicator for assessing the reliability of the\n reconstructions in the absence of the ground truth. We demonstrate the applicability of the current\nKeywords:\n methods on a tantalum sample.Microstructure reconstruction\nConvolutional neural network \u00a9 2019 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.\nElectron backscatter diffraction\n\n\n\n\n\n1. Introduction characterization of relatively large (0.1e10 mm) polycrystalline\n samples [6,7]. EBSD provides orientation maps at sub-grain spatial\n Spatial microstructural heterogeneities at the mesoscale resolution (~200 nm) and high crystal orientation resolution (<0.5\n(~1e100 mm), including grain boundaries, phase boundaries, dis- deg). During an EBSD measurement, an electron beam is focused\nlocations, defect structures, and chemical segregation, directly onto a spot on the sample surface and a diffraction pattern is\ncontrol macroscopic material properties and behavior. Currently, recorded. Top of the line commercially available orientation imag-\nthe lack of understanding of the complex collective response of ing microscopes (OIM) can make one such measurement in less\nmesoscale features to external stimuli poses fundamental chal- than 1 ms [8]. When optimizing for speed, the pattern quality de-\nlenges in predicting materials\u2019 microstructure evolution [1,2]. This grades, which results in reduced orientation indexing \ufb01delity with\nis critical for engineering new materials with tailored properties. conventional reconstruction methods [9]. In contrast, the state-of-\nExperimental techniques have advanced signi\ufb01cantly in the last the-art dictionary indexing (DI) method signi\ufb01cantly improves the\ntwo decades, enabling destructive and non-destructive character- orientation reconstruction accuracy from highly noisy EBSD pat-\nization of material microstructures and their evolution at high terns [9,10]. However, real-time orientation indexing using the\nspatial and temporal resolutions [3e5]. With increasing data dictionary learning method is not yet viable, unable to match scan\ncollection rates and sizes of datasets per sample reaching tens of speeds which can reach as high as 1 ms/pattern, limiting the DI\nTBs, there is considerable interest in exploring how ef\ufb01ciently approach to be an off-line method [11,12].\nanalysis methods based on machine learning can be applied to the Fast orientation indexing from noisy patterns is crucial for in-\nreconstruction process. In this work, we show how convolution situ microstructure evolution [7,9] and other quantitative studies\nneural networks (CNNs) can be used as \u201csurrogate\u201d models for that require statistically signi\ufb01cant volume measurements at high\nsolving the inverse problem of reconstructing microstructures from spatial resolution [3]. In spite of progress towards fast data collec-\ndiffraction data. tion times and computation, there is a need for new methods to\n Electron backscatter diffraction (EBSD) is the most extensively accelerate the orientation indexing from noisy EBSD patterns,\nused technique for 2D and/or 3D spatially resolved crystallographic especially if we are to make progress towards real-time in-situ\n studies and 3D microstructure characterization.\n Recently, deep convolutional networks have been utilized for a\n * Corresponding author. range of imaging techniques from automated image segmentation\n E-mail address: reeju@lanl.gov (R. Pokharel).\n\n\nhttps://doi.org/10.1016/j.actamat.2019.03.026\n1359-6454/\u00a9 2019 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131 119\n\n\n[13,14] to image reconstruction for magnetic resonance imaging high rates.\n[15]. While CNNs have demonstrated immense success for prob-\nlems related to image classi\ufb01cation [16] and detection [17], only a 1.2. Outline\nfew studies have utilized CNNs to address regression problems\n[18e20] for improving the performance of a forward model. The basic principle of EBSD (measurements and simulations) is\n Liu et al. [11] demonstrated the \ufb01rst application of convolu- presented in Section 2. A brief background on CNNs is reviewed in\ntional neural networks for EBSD orientation indexing from simu- Section 3, and their application to EBSD is presented in Section 4. In\nlated diffraction patterns. Separate CNN models were trained to Section 5, we summarize our key \ufb01ndings related to the model's\npredict individual components of the crystal orientation. Jha et al. sensitivity to image noise and experimental parameters, and\n[12] extended this approach to predict the full crystal orientation discuss the computational cost of the proposed methods.\nusing a deep CNN network, employing mean disorientation in the\nloss function to account for the non-linearity of the orientation 2. Electron backscatter diffraction\nspace. However, the implementation of complicated loss functions\nled to relatively long training times. In both the studies, the models A diffraction pattern (intensity distribution) on a detector, X \u00bc\nwere only trained and tested on simulated EBSD patterns, in X\u00f0x;y\u00de, where \u00f0x; y\u00de is pixel location, can be expressed as a function\ncontrast to real experimental diffraction patterns; therefore, the of crystal orientation represented as a quaternion q \u00bc \u00f0q0;q1;q2;q3\u00de,\neffects of measurement noise were not considered. In this work, of the form:\nwe demonstrate the ef\ufb01cacy of the neural network approach for\nexperimentally measured data with varying noise levels. We pre- X \u00bc P \u00f0f\u00f0qjM ; C ; G \u00de\u00de; (1)\nsent two convolutional neural networks based EBSD reconstruc-\ntion techniques that are applicable to experimental diffraction where M parameterizes the microscope, C denotes the crystal-\npatterns and robust towards measurement noise. In particular, we lography of the sample, G represent the geometry of the setup, and\nshow that by utilizing a simulation-trained CNN to provide an P \u00f0l\u00de represents a random Poisson sample drawn from a Poisson\ninitial estimate for re-training a small part of the network or distribution with mean l. Like many other diffraction-based mea-\ncombining it with an additional network with a small amount of surement methods, reconstructing a crystallographic orientation\nexperimental data results in high prediction accuracy for from an electron backscatter diffraction (EBSD) pattern is achieved\nmeasured data. by solving an inverse problem, i.e. \ufb01nding the inverse function\n which maps a g\n\n## 5. shen2019convolutionalneuralnetworkbased pages 8-10 (Context ID: pqac-00000016)\n\n prediction. A correction for the orienta- mental diffraction patterns of the same orientation collected at\ntion error induced by the pattern center shift across a large area various exposure times. The top row is the experimentally\nEBSD scan can be applied using the method described in Singh et al. measured patterns and the bottom row corresponds to the\n[34]. While we do not use this correction factor here, as we are FusionNet transformed patterns. The measured pattern quality\nprimarily interested in demonstrating that large reconstruction signi\ufb01cantly degrades with decreasing exposure times. For shorter\nerrors can be reduced through re-training (shown in Fig. 7), ac- exposures, the reconstructed patterns are more deformed, but\ncounting for the pattern center shift as a function of scan position many features are retained in the transformed image. Fig. 8(b)\nmay further reduce the low misorientations shown in Fig. 7(f). shows the simulated pattern corresponding to the same orientation126 Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131\n\n\n\n\n\nFig. 7. (a), (d) The misorientation map between the predicted and the ground truth crystal orientations. The red points correspond to the positions from which measured diffraction\npattern and crystal orientation pairs were used for training the U-Net and the re-trained networks. (c), (f) Misorientation distribution calculated for the 100 measured patterns,\nrandomly chosen from the right most side of the sample (unseen during training), used as a testing set for the two models. The measured diffraction patterns from the remaining Ta\nsample positions were used for orientation prediction. Low misorientation angle indicates good agreement. (b), (e) Spatial map for the deviation from unity of the quaternion norm.\nLarger deviation indicates higher prediction error. (For interpretation of the references to color in this \ufb01gure legend, the reader is referred to the Web version of this article.)\n\n\n\n\n\nFig. 8. (a) EBSD detector patterns. The \ufb01rst row shows the measured diffraction images of the same orientation collected at various exposure times. The second row shows the\nFusionNet transformed images that were used as the input for the CNN model for orientation indexing. (b) The simulated pattern corresponding to the same orientation in (a). (c)\nThe distributions of misorientations between the predicted and the ground truth crystal orientations of the 601 test dataset collected at various exposures. (d) Average misori-\nentation angles for data points with less than 10+ misorientation versus the exposure times for CNN and Hough transform analyses. The error bar is the standard deviation obtained\nfrom the misorientation distribution. Both methods exhibit a decrease in accuracy with increasing noise level. Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131 127\n\nin Fig. 8(a). Misorientation distributions between the measured and pixels (pixel size is 71.4 mm). This was expected because detector\nthe predicted orientations for the entire scanned area are shown in translation and crystal rotation have approximately the same effect\nFig. 8(c). The results demonstrate that the CNN approach can, to a on the EBSD pattern. Therefore, similar to Hough transform and\ndegree, learn to index very low-quality diffraction images. dictionary methods, the proposed CNN based approach also re-\n A comparison between the CNN and Hough transform recon- quires determination of the experimental setup parameters before\nstruction accuracy is shown in Fig. 8(d). Here, we show what per- reconstruction. Therefore, a single neural network will not provide\ncentage of points have less than 10+ misorientations with the accurate predictions for all experimental geometry parameters;\nground truth, as shown by the gray bars. Note that the CNN however, it might be possible to account for the change in instru-\nconsistently achieved a higher percentage of such points for all ment parameters via re-training the weights of the previously\nexposure times. The difference is especially large for 1 ms exposure trained CNN model.\ntimes where the CNN approach resulted with 50% of all points\nhaving < 10+ misorientations, while the Hough transform method 5.4. Computational cost\nonly had 15% of the points with < 10+ misorientation. Next, us-\ning only these < 10+ misorientation points, we calculated the Compared to a dictionary approach, one advantage of using\naverage misorientation and the standard deviation, as shown in neural networks is increased computational performance.\nblue. For this choice of points, the Hough transform method has Reducing computation time is an important consideration in\nlower average misorientation and lower variance. Overall, the CNN designing a network architecture. As we can see from Equation\napproach is more robust to noise because a larger number of points (A.1), the computation time for each pixel in the convolutional layer\nhave overall lower predicted misorientations, while the Hough is proportional to the kernel size and the number of \ufb01lters, which is\ntransform method is more accurate but only for a very small frac- related to the number of computations performed by the network.\ntion of points, suggesting that only points with low noise were Thus, reducing the number of layers and keeping the kernel size\naccurately reconstructed. and \ufb01lter numbers managable for the convolutional layers will\n Recent studies have also demonstrated CNNs\u2019 robustness to increase the performance. However, a CNNs\u2019 ability to approximate\nnoisy data [35,36]. This is not surprising because convolutional complex functions depends on the network size [41,42], and the\nlayers have been shown to be capable of learning the features from prediction power will decrease with decreasing network size.\nthe training data to recover a clean image from a noisy observation The network architectures of the CNN and U-Net used in this\n[37e39]. Moreover, CNNs can learn high level features automati- work were optimized for both computational performance and\ncally and are very robust to local noise, which makes them prediction accuracy, and we demonstrated that the models gener-\noutperform traditional methods based on handcrafted features, alized well for both unseen simulated and experimental datasets.\nwhich involve pre-determining characteristic regions or de- Using the network structure described here, the U-Net method\nscriptors of an image, utilized in many computer vision tasks took about 17 min to reconstruct the 1 1 mm2 scanned region\n[15e17,40]. (251001 data points) on the sample (i.e. 4 ms per pattern) on a\n MacBook Pro with a 2.8 GHz Intel Core i7 processor, and the\n5.3. Effects of uncertainty in experimental parameters retrained CNN took about 5 min (i.e. 1 ms per pattern). Further-\n more, there are many neural network frameworks (e.g. tensor\ufb02ow\n Another challenge always encountered in experiments is the [43]) that can utilize GPUs to further speed up the reconstruction\nuncertainty in experimental setup parameters. We conducted a without changing the code.\nparametric study to investigate the effect of instrument setup un-\ncertainty on model prediction. For that purpose, we trained our 5.5. Combining CNN with the dictionary approach\nmodel with simulated images generated from one set of detector\nparameters, and tested it on the images that were generated using Both Liu et al. [11] and Jha et al. [12] reported the dictionary\ndifferent setup parameters. The results are shown in Fig. 9. In indexing approach to be slower than the CNN method. In this work,\nFig. 9(a), we only changed the distance between the detector and we have demonstrated that the CNN-based method is relatively fast\nthe sample, keeping everything else \ufb01xed. As long as the distance and can enable real-time reconstructions during the scanning\noffset was smaller than 900 mm, the results were not strongly process. However, the CNN method is not as accurate as the dic-\nin\ufb02uenced. Fig. 9(b) and (c) show the effects of X and Y translation, tionary approach, which can reconstruct noisy data with high ac-\nwhich caused the error to shift towards the right suggesting poor curacy [10]. Further improvements may be made by combining the\nprediction. The extent of this shifting was dependent on the choice two approaches: \ufb01rst apply CNN-based regression to obtain pre-\nof the experimenta\n\n## 6. shen2019convolutionalneuralnetworkbased pages 10-12 (Context ID: pqac-00000017)\n\n we have demonstrated that the CNN-based method is relatively fast\nthe sample, keeping everything else \ufb01xed. As long as the distance and can enable real-time reconstructions during the scanning\noffset was smaller than 900 mm, the results were not strongly process. However, the CNN method is not as accurate as the dic-\nin\ufb02uenced. Fig. 9(b) and (c) show the effects of X and Y translation, tionary approach, which can reconstruct noisy data with high ac-\nwhich caused the error to shift towards the right suggesting poor curacy [10]. Further improvements may be made by combining the\nprediction. The extent of this shifting was dependent on the choice two approaches: \ufb01rst apply CNN-based regression to obtain pre-\nof the experimental setup, and in our case, it was about 4+ per 20 liminary predictions and a measure of con\ufb01dence for each location\n\n\n\n\n\nFig. 9. The misorientation distributions when the training and testing patterns are generated under different detector parameter setups. (a) The sample to detector distance is\ndifferent for training and testing data, where the offset unit is mm. (b) The pattern center X coordinate on detector are different for training and testing data. (c) The pattern center Y\ncoordinate on detector are different for training and testing data. The offset units for X and Y are given in number of pixels (pixel size is 71.4 mm).128 Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131\n\n\ngiven by the norm of the predicted quaternion. For locations with methods improved the accuracy of the experimental data recon-\nhigh con\ufb01dence reconstructions, run the pattern matching only in struction. This improvement can be understood by looking at the\nthe neighborhood (e.g. 10+) of the preliminary results in the dic- activation map of the last convolutional layer, which after max\ntionary. For those with low con\ufb01dence, run the full dictionary pooling is the input to the fully connected layer.\nsearch. Fig. 10 shows one such example from the test dataset. Four\n feature maps are shown, one for each of the four \ufb01lters in the layer.\n5.6. Regularization The \ufb01rst row is the simulation image and the \ufb01lter outputs of the\n original CNN. We can see that some features related to Kikuchi\n Regularization is important for reducing the chance of over- bands have been learned. The second shows the results of inputing\n\ufb01tting. We used L2 regularizations for our models without using experimental data into the CNN model which has been trained on\ndropout [44]. Dropout is a common regularization technique used simulations only. Clearly the original CNN will respond differently\nin most CNN classi\ufb01cation problems. It randomly forces the neuron to experimental and simulation patterns. The difference was sig-\nactivations to be zero in training, but does nothing in testing. Pre- ni\ufb01cant enough that the original CNN was unable to accurately\nvious studies have shown that dropout can improve the perfor- predict quaternions for experimental patterns.\nmance of a particular CNN-based regression model. For example, The third row shows the \ufb01lter outputs after a U-Net transformed\nKendall et. al. [45] used dropout in their model to infer camera image is fed into the original CNN; the activation maps look very\npositions from a picture taken by that camera, and saw improved similar to those shown in the \ufb01rst row, which were generated from\npredictions. We tried two dropout ratios (0:3; 0:5) in our CNN the simulated image. In the fourth row, we retrained the CNN so\nmodels, but they resulted in lower \ufb01delity predictions. However, that the weights of convolutional layers were adapted to the\nwithout dropout, we obtained validation errors that were already experimental images. We can see that the activation maps are also\nas low as the training error, as seen in Fig. 4(a). The reconstruction very similar to those in the \ufb01rst row, which supports our assump-\naccuracy (mean error and variance of error) of our model for un- tion that differences between simulation and experimental pat-\nseen data and for training data was similar, indicating that over- terns are low level features and they can be learned solely by\n\ufb01tting was not taking place. convolutional layers.\n The two approaches presented here for experimental pattern\n5.7. Microstructure reconstruction indexing are simple and robust. Similar studies have being con-\n ducted on transfer learning and domain transforms [46]. One such\n Experimental EBSD data reconstruction was performed by example is a generative adversarial networks (GAN) [47] that is\nemploying two adaptive learning networks. One of the reasons for utilized to transform images from one domain to another [48,49].\nnot training another CNN model from scratch was due to the lack of Essentially, the generative networks are learning the metric for\nlabeled experimental patterns that spanned the entire orientation measuring similarity between images produced from different mo-\nspace (fundamental zone) for a given material crystal structure. We dalities, which is similar to our approach utilizing the U-Net and\nobserved that both the domain transform and transfer learning FusionNet to transform experimental patterns to simulated patterns.\n\n\n\n\n\nFig. 10. Activation maps for the last convolutional layer. Case 1: simulated pattern as input to the original CNN. Case 2: experimental pattern as an input to the original CNN. Case 3:\nU-Net transformed pattern as an input to the original CNN. Case 4: experimental pattern as input to the re-trained CNN. Y.-F. Shen et al. / Acta Materialia 170 (2019) 118e131 129\n\n\n6. Conclusions\n\n This work demonstrates the effectiveness of two CNN ap- Xs Xs X 1 hl\u00f0i;j\u00de;c \u00bc w\u00f0m;n\u00de;c;c0al\u00f0i m;j n\u00de;c \u00fe bc;c0 (A.1)proaches for microstructure reconstruction from measured EBSD\n m\u00bc s n\u00bc s c0\npatterns. Both approaches initially trained the CNN with a large\nnumber of easily generated simulation data and then utilized a which is followed by the activation function\nrelatively small number of expensive experimentally measured\npatterns to either re-train several convolutional layers or transform\nexperimental patterns into simulated patterns. Both methods were n o\n al\u00f0i;j\u00de;c \u00bc ReLU hl\u00f0i;j\u00de;c \u00bc max 0; hl\u00f0i;j\u00de;c ;applied to a tantalum sample with a body centered cubic crystal\nstructure. Noise in the patterns was one of the major contributors\nthat impacted the model prediction signi\ufb01cantly. The proposed where hl\u00f0i;j\u00de;c and al\u00f0i;j\u00de;c are the feature map intensity and activation\nCNNs demonstrated robustness to varying levels of noise in real value at \u00f0i; j\u00de pixel coordinates, s is the kernel size and c is the\nexperimentally measured data (achieved by changing exposure\n channel index in layer l. The w\u00f0m;n\u00de;c0;c are the convolutional kernel\ntimes during data collection) and successfully reconstructed the\n weights between channel c and channel c0 of neighboring layers\norientations while matching the highest EBSD scanning rates. The\n and bc;c0 is the bias term.\nproposed quaternion norm as a measure of con\ufb01dence was\n We used a 3 3 kernel for all convolutional layers with s \u00bc 1,\ndemonstrated to be a strong indicator for assessing the reliability of\n the range of m and n was m; n2f 1; 0; 1g, and zero padding was\nthe reconstructions. This will be crucial for determining the quality\n employed. The \ufb01lter numbers (i.e. number of channels) were 4; 8; 8\nof reconstructions for real applications, when the ground truth is\n \n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013, pqac-00000014, pqac-00000015, pqac-00000016, pqac-00000017]\n\nStatus: Paper Count=25 | Relevant Papers=8 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=18 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_4klcSpImEsXAZeSqWkaHiygh", + "info": { + "start_ts": 755.581905948, + "end_ts": 760.816464142 + } + }, + { + "role": "tool", + "content": "## 1. sparks20243dreconstructionof pages 23-24 (Context ID: pqac-00000000)\n\nages and contrast-enhanced\nin the optical images, they were histogram-matched using BSE images and subsequently applied to the raw images\nScikit-Image\u2019s \u201cmatch_histograms\u201d function, with the first and binary mask data types. As the first layer of the optical\noptical image serving as the reference histogram for the later data serves as the reference location of the sample\u2019s spa-\nlayers to match against. From there, the images were spa- tial information, the optical data were centered in physical\ntially registered in series using OpenCV\u2019s implementation space around the center point of the first layer\u2019s binary mask\nof the SIFT algorithm. The default parameters were used, using SITK\u2019s \u201cCenteredTransformInitializer\u201d function. The\nand feature matching was done using OpenCV\u2019s brute force centering transform from the first optical layer was then\nmatcher, keeping the top 30% of matched features for the cal- applied to each subsequent layer to ensure that any global\nculation of the transformation matrix using \u201ccv2.estimateAff- alignment shifts due to shear or sample rotation captured\ninePartial2D.\u201d This function calculates a rigid transform, in the SIFT registration of the optical images were carried\nonly allowing for rotation and translation. After alignment, through to the BSE and EBSD registration steps.\na binary mask of the sample was automatically found using The EBSD data were then registered against the updated\na combination of Otsu thresholding, binary hole filling, and BSE images using the same SITK framework, with the\nsegmenting out the binary feature with the largest area. This B-Spline mesh size being updated to 3x3 as determined in\nbinary mask is used to mask out the mounting medium (Poly- [2]. Grayscale conversions of EBSD inverse pole figure (IPF)\nfast) and the gold cube while registering the BSE data against colormaps were registered against the contrast-enhanced\nthe optical data, as well as determining the sample center for BSE images, allowing for the nonlinear registration of grain\nlater alignment with the BSE and EBSD data. boundary information to remove the distortions from EBSD\n The same general SITK multi-resolution image registra- data collection. The IPF colormaps were generated as part of\ntion framework as outlined in [2] was used for the BSE to a Python script using ORIX, an open-source Python EBSD\noptical and EBSD to BSE registrations, with differences in analysis suite [47, 48]. ORIX\u2019s \u201corientation2color\u201d func-\nresolution and hyperparameters as needed. This framework tion was used to generate the IPF colormaps for each layer,\nallows for intra- and inter-modal image registration using while binary masks for the EBSD data were calculated using\na variety of different similarity metrics and transforma- the Otsu threshold method on the confidence index values\ntion types. The parameters used were as follows: B-Spline from the EBSD scan data. The BSE images were down-\ntransform, mutual information as the similarity metric [46], sampled from a 2500\u00d72500 grid with a spacing of 0.5 \u03bcm/\nL-BFGS-B as the optimizer, and nearest-neighbor interpola- pixel to a 600\u00d7600 grid with a spacing of 2 \u03bcm/pixel using\ntion for final image resampling. As SITK image objects are SITK\u2019s \u201cResample\u201d function to match the SITK grid used\narray-based images defined on a physical-space region, each for processing the EBSD data. Note that both the BSE and\ndata mode\u2019s image is provided along with its physical grid EBSD grids in SITK were larger than the original images\nspacing ( \u03bcm/pixel). This allows for data modes at different to allow for motion during registration without loss of data;\nresolutions to be registered, as well as the easy resampling added \u201cpadding\u201d pixels near the boundaries of the grids\nof higher-resolution data modes down to the lowest resolu- were filled with zero values to match the dark background\ntion reference grid. of both images. The B-Spline transform was then calculated\n The BSE images were registered against the aligned on a per-layer basis and applied to the raw EBSD orientation\noptical data using an updated version of the Python script data files with nearest-neighbor interpolation using a cus-\nfrom [2]. One difference from the earlier study is that only tom Python input\u2013output function. Any orientations in the\nthe BSE images and not the optical images were contrast- exported EBSD data corresponding to padding pixels were\nenhanced, as contrast enhancement of the optical images set to the identity rotation with a confidence index of zero.\nfrom this study emphasized features such as surface Separately from the SEM data modes, the SWLI data\nscratches that were not shared with the BSE images and were also registered against the self-aligned optical images.\nresulted in a poor nonlinear registration. The BSE images The raw heightmap data were corrected to remove any tilt\nwere treated with Scikit-Image\u2019s \u201cequalize_adapthist\u201d func- present from data collection, which was done by calculating\ntion to perform adaptive histogram equalization on each the surface plane of the sample using least squares estima-\nlayer, followed by a \u201cmatch_histograms\u201d treatment, match- tion and subtracting the resulting plane from the raw data.\ning each sequential layer with the first BSE layer to improve The tilt-corrected heightmaps were then registered against\ngrayscale consistency across the image series. Binary masks the optical images using the same SITK framework and\nfor the BSE images were then calculated using the same subsequently downsampled to a 600\u00d7600 grid with 2 \u03bc\u200am\nmethod as the optical images. spacing to match the grid used for the corrected EBSD data,\n For the BSE to optical registration, the B-Spline ensuring that each height value corresponded to a single\nparameters were a mesh size of 1x1 and a transformation EBSD data point. Figure 6 (d) and (h) shows an example of\norder of 2. The B-Spline transform was calculated using the raw and tilt-corrected heightmaps.796 Integrating Materials and Manufacturing Innovation (2024) 13:773\u2013803\n\n\n\n\n\nFig.\u202f26\u2002 \u2009SEM micrographs for section 331 showing the damaged region in the top left corner of the sample\n\nAppendix G.1: Registration Challenges are usually given as distances from the origin of a chosen\n reference frame, while crystal orientations are generally\nChallenges with the B-Spline registration of EBSD to BSE specified as rotations relative to an \u201corigin\u201d reference frame.\ndata occurred when one corner of the sample was accidentally It is possible for the reference frame used to define the posi-\ndamaged within the mount partway through the serial section- tion values (the spatial reference frame) to be different than\ning series. As the B-Spline transform and mutual information the reference frame used to define the rotations represent-\nsimilarity metric operate on a pixel-by-pixel basis, the SITK ing orientations (the grain reference frame), so both must\noptimizer distorts the moving image as needed to maximize be considered separately. The recorded rotations may also\nthe metric score. For the damaged region, there is no longer a differ in one or more choices required when specifying a\nclear grain structure or much shared information between the rotation in numerical form [49], such as using either the\nEBSD and BSE data types, causing the optimizer to incorrectly active or passive rotation convention. As all of these choices\ndistort the damaged region with respect to the rest of the sam- frequently vary between different data formats, cross-format\nple. In some areas of the sample, the poor distortion correc- comparisons require significant attention to these details to\ntion from the damaged region causes neighboring grains to be ensure that any differences have been accounted for.\npulled away from their true locations as well. Overall, the data We employ MTEX [37] to manipulate the EBSD and\nin the damaged region are not usable, and attempts were made NF-HEDM orientation maps, as it has native support\nto minimize its impact on the nearby regions by thresholding out for many existing orientation data formats and can be\nlow-confidence EBSD data, or potentially masking out the entire extended as necessary to accommodate new ones. The\ndamaged region from the registration optimization. Figure 26a (\u200axU\u200a, yU\u200a, zU\u200a) unified coordinate system shown in Fig. 8\nand Fig. 26b show the BSE and EBSD scans on section 331, is also the one used in MTEX, with the origin in the\nat the start of the damaged region, as well as the impact of the upper left corner, xU pointing to the righ\n\n## 2. sparks20243dreconstructionof pages 9-12 (Context ID: pqac-00000001)\n\nruns.\nagonal reconstruction grid onto the square 3D-EBSD posi-\ntion grid to allow for point-to-point quantitative compari- Reference Frame Misalignments Between Different Layers\nsons. Here, we use the coordinate system employed by the of the Same Type\n3D-EBSD position data as the unified coordinate system and\nbring the HEDM data into this coordinate system, as shown To estimate the rigid-body misalignment between two differ-\nin Fig. 8. Subsequent maps use this coordinate system. ent layers of the same type, we first calculate the misorienta-\n tions between the two layers at each (\u200axU\u200a, yU\u200a) point, similarly\n to Fig. 9. Any points with misorientation angles above the\nResults and Discussion 3\u200a\u25e6 threshold are discarded, and the remaining point-to-point\n misorientations are then averaged to produce a mean mis-\nA detailed catalog of the data is tabulated and included in orientation between the two different layers. These layer-\nthe dataset archived at the MDF. In this section, we present to-layer misorientations can be averaged further to obtain\nseveral comparison studies conducted using the 3D orienta- a mean misorientation between one layer and every other\ntion map of the VOI generated from MPSS and NF-HEDM. layer, e.g., averaging the 19 results from layer 1 of the NF-\n HEDM data versus layers 2 through 20. Averaging is done in\nLayer\u2011to\u2011Layer Orientation Comparisons quaternion form to account for all misorientation parameters\n [34], although the discussion below focuses on the misori-\nOnce the EBSD and NF-HEDM maps have been registered entation angles.\nand resampled onto the unified coordinate system, they can For the reconstructed NF-HEDM data, the misorien-\nbe easily compared across corresponding points, whether tations are uniformly very small, with all layer-to-layer\nthose are from a different data type (i.e., comparing points misorientation angles below 0.05\u25e6\u200a. As shown in Fig. 10,\nwith matching (\u200axU\u200a, yU\u200a, zU\u200a) values across data types) or a dif- higher misorientation angles are found between individual\nferent layer in the same data type (i.e., comparing points with sections of EBSD data due to the sample needing to be\nmatching (\u200axU\u200a, yU\u200a) values but different zU values within a sin- removed from the SEM for mechanical polishing and the\ngle data type). In this work, a misorientation angle threshold SEM stage needing to be moved between 0\u200a\u25e6 and 70\u25e6 tilt\nof 3\u200a\u25e6 is used to define whether a pair of points belong to the for every section. Despite the higher variation between\nsame grain in two compared datasets. Figure 9 shows that sections for the EBSD data, the mean misorientation\nthe observed misorientations between EBSD sections are angle between one section and all other sections never\ngenerally either well above or well below this threshold, with exceeds 0.3\u25e6\u200a, while the individual section-to-section mis-\nthe number of points above the threshold increasing as the orientation angles never exceed 0.5\u25e6\u200a, both low enough to782 Integrating Materials and Manufacturing Innovation (2024) 13:773\u2013803\n\n\n\n\n\nFig.\u202f9\u2002 \u2009Section-to-section misorientation maps from the distortion-corrected EBSD data\n\n\n\n\n\nFig.\u202f10\u2002 \u2009Mean misorientation angles between the distortion-corrected EBSD data for a given section and either the next section or all other sec-\ntions, and an equivalent plot for the NF-HEDM data layers\n\nbe below the noise threshold of conventional EBSD. We relative to comparisons of adjacent layers within a single\nattribute this high consistency to the combination of high- data type (see Fig. 11 versus Fig. 9). Across a subset of\nprecision EBSD, pattern center fitting for every section, the VOI cropped to contain only the sample, the fraction\nautomated sample alignment routines to ensure that the of noncorresponding points (those with a misorientation\nsample is consistently positioned during SEM imaging angle of more than 3\u200a\u25e6\u200a) was found to be 46.2%. By com-\nand EBSD scanning, and an effective clamping solution paring the IPF colormaps of both data types (as shown\nto ensure that the sample mounting on the SEM stage is in Fig. 8), we found that these points typically involved\nhighly repeatable. regions where smaller grains or twins visible in the EBSD\n data were absent from the HEDM data, with the larger\nMisorientations Between Resampled EBSD and NF\u2011HEDM grains found in the HEDM reconstruction having expanded\nData Layers to cover those regions. This result is similar to the findings\n of previous EBSD/HEDM comparison studies [1\u20134], with\nComparisons between matching layers of the resampled improving the correspondence of reconstructed HEDM\nEBSD and NF-HEDM data were similar to layer-to-layer data to the ground truth being a critical research target.\ncomparisons within a single data type, but a significantly In addition to the network of noncorresponding points\nhigher fraction of noncorresponding pixels was observed near the outlines of larger grains, Fig. 11 also shows aIntegrating Materials and Manufacturing Innovation (2024) 13:773\u2013803 783\n\n\n\n\n\nFig.\u202f11\u2002 \u2009Maps of misorientation angle and misorientation axis components versus position for the first pair of resampled 3D-EBSD and NF-\nHEDM data layers\n\n\nrelatively consistent misorientation axis and angle in the any individual layer\u2019s alignment rotation and the average\nregions that do correspond. Further investigation showed alignment rotation across all layers was 0.21\u25e6\u200a, similar in\nthat this was true for all resampled EBSD layers when com- magnitude to the misorientation between EBSD sections\npared with their matching HEDM layers. This prompted the shown in Fig. 10 (as might be expected given the high inter-\nidea of applying a per-layer uniform rotation to the EBSD nal consistency of the HEDM data).\norientation data to better align it with the HEDM orienta- The source of this misorientation between the EBSD and\ntion data, empirically determined from the average misori- HEDM orientation data has not been confirmed, but the\nentation axis and angle for corresponding points in that pair high level of consistency across the entire dataset implies\nof layers. This was highly successful at reducing the mis- that it is a systematic error. For the EBSD analysis, plau-\norientation between the EBSD and HEDM data, with 95% sible sources of systematic error include the true EBSD\nof the corresponding points after such alignment having a camera tilt angle differing from its nominal value of 10\u25e6 or\nmisorientation angle below 0.46\u25e6 across the entire volume the true sample tilt in the EBSD position differing from its\nof interest, with a mean value of only 0.24\u25e6\u200a. While these nominal value of 70\u25e6\u200a. Either of these is known to produce\nalignment rotations were calculated per layer (allowing for a fictitious rotation of the crystal orientations about xU in\nthe known variations between EBSD layers to be accounted direct proportion to the difference between the nominal\nfor), the alignment rotations themselves were also remark- angle and the true angle. However, it is unlikely that the\nably consistent from layer to layer. The largest angle between latter would produce such a consistent error across multiple784 Integrating Materials and Manufacturing Innovation (2024) 13:773\u2013803\n\n\n\n\n\nFig.\u202f12\u2002 \u2009Misorientation angle between corresponding points in the ing or Hough indexing on the same raw EBSD patterns, averaged\nresampled EBSD and HEDM data (after per-layer empirical orienta- across zU and plotted versus (\u200axU\u200a, yU\u200a) position. Note the significantly\ntion alignment rotations) when using either spherical harmonic index- different color scales for the two plots\n\nsections, as the SEM stage is moved for every section. The PC shift calibration in the Y-direction was not correctly\nSEM stage tilt axis could also be slightly misaligned rela- adjusted for the large scan area used and might provide\ntive to the EBSD camera, which is known to cause consist- better performance if this was rectified. However, even in\nent errors in the measured orientations from EBSD [35]. the central region where the PC shift effec\n\n## 3. stinville2022multimodaldatasetof pages 9-12 (Context ID: pqac-00000002)\n\nlements produced within grain surface meshes, but algorithm unaware of grain ID assignment.\n (c) Final mesh after segmenting elements according to grain IDs.\n\n\n clusters of the same material connecting at a single voxel corner) that could create singularities in the finite\n element solution. The resulting geometric model represents each grain as a region (volume) with geometric\n faces (surfaces) representing grain boundaries. Attributes attached to each region allow the user to retrieve the\n grain ID as it was defined in the originating DREAM3D15 dataset. At this stage, the face geometry still reflects\n the stair-stepped boundaries between the individual voxels, therefore a geometric based algorithm is used to\n create smooth geometric faces while preserving the overall shape of the grain boundaries. The resulting geomet-\n ric model can be tagged with meshing and analysis attributes to generate a run-ready input deck for the finite\n element solver.\n For the In718 RVE, features smaller than 50 connected voxels were removed followed by an erosion/dilation\n step using a 3x3x3 block structuring element. Many of the grains in this dataset were rather thin (often only one\n or two voxels thick). These were not processed using the erosion/dilation algorithm since it would completely\n remove them from the data. Figure 13 shows a twin grain (identified as grain ID 105 in the DREAM3D dataset)\n with a cut plane showing that the grain is at times only one voxel thick. Figure 14 shows pictures of that grain as\n it goes through the reconstruction and meshing process, from the original voxel dataset to the voxel structure\n after artifacts were removed, followed by the smoothed grain surface representation and the associated volu-\n metric mesh. Figure 15(a) shows a view of the In718 RVE geometric model that was constructed from the voxel\n dataset. The corresponding mesh shown in Fig. 15(b) was created by assigning a mesh size attribute with value\n 0.3 \u03bcm on all grain boundaries, resulting in a total of 16.7 million tetrahedral elements. A close-up view of the\n\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 9www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n\n\n\n Fig. 13\u2002 Cut plane through the grain number 105. Twin grains are as thin as one voxel at times within the\n voxelized 3D dataset.\n\n\n\n\n\n Fig. 14\u2002 Grain number 105 as it passes through the modeling process. (a) Initial voxel-level grain representation.\n (b) the voxel data after artifacts were removed. (c) geometric representation of the grain and, (d) the corresponding\n discretization.\n\n\n\n\n\n Fig. 15\u2002 Modeling procedure using Simmetrix\u2019 software suite. (a) Geometric representation of the In718 RVE\n created from the serial-sectioning voxel dataset. (a) Volumetric discretization of the In718 RVE.\n\n\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 10www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n Load step E1 E2 E3 E4\n Macroscopic strain 0.69% 0.91% 1.23% 1.83%\n Average \u03b5xx strain from HR-DIC 0.68% (1.5%) 0.87% (4.4%) 1.16% (5.7%) 1.70% (7.1%)\n\n Table 1.\u2002 Comparison between macroscopic total strain and average \u03b5xx from the HR-DIC measurements.\n The absolute difference between average \u03b5xx from HR-DIC measurements and macroscopic strain is given in\n parentheses.\n\n\n\n\n\n Fig. 16\u2002 Close-up view of the RVE representation. (a) Voxelized 3D dataset. (b) Geometric model. (c)\n Volumetric mesh.\n\n\n In718 RVE model evolution from a voxel level representation to a geometric reconstruction and an associated\n volumetric mesh is provided in Fig. 16.\n Data Records\n SEM Images.\u2003 The raw SEM images used for HR-DIC calculations are 16-bit tiff images and can be down-\n loaded from the DRYAD repository29. The nomenclature and location of the images are detailed in Fig. 2. The\n image origin is located for the pixels within the tiff images, which is set as the first top-left pixel, with the x-tiff\n (Xg) proceeding to the right, and y-tiff (Zg) proceeding down the image.\n Full-field measurements from HR-DIC.\u2003 The HR-DIC data are provided as 32-bit tiff images. The images\n can be viewed using software such as ImageJ. The values at each pixel in the images are the quantitative physical\n value of the displayed full fields given in unit length for the strain field \u03b5xx, in pixels for the in-plane slip amplitude,\n and in radians for the in-plane slip direction. We refer the reader to10,12 for more information about the physical\n meaning of these quantities. The data can be downloaded from the DRYAD repository29. Adjacent pixels in all\n maps represent a physical distance of 100.34 nm. The values in the in-plane slip amplitude maps are given in pix-\n els, where one pixel represents a shearing/displacement induced by slip of 33.45 nm. The image origin is located\n for the pixels within the tiff images, which is set as the first top-left pixel, with the x-tiff (Xg) proceeding to the\n right, and y-tiff (Zg) proceeding down the image.\n 2D Crystallographic orientation data.\u2003 The IPF maps along the loading direction X (see Fig. 4) from the raw\n EBSD measurements and after cleaning and distortion (to match the DIC full field maps) are provided in a tiff format\n in a RGB color type and labeled as IPF_Raw X.tif and IPF_DistordedDIC_X.tif, respectively. The IPF map along the\n transverse direction Zg (see Fig. 4) is also provided and labeled as IPF_Raw_Z.tif. The raw crystallographic orientation\n data (Euler angle and grain structure) are provided in the .osc and .ang format and labeled as Euler_Orientation_Raw.\n osc and Euler_Orientation_Raw.ang. The .osc and .ang format are labeled as Euler Orientation Raw.osc and Euler\n Orientation Raw.ang. The .osc format can be opened using the TSL software from EDAX. The .ang format is a ASCII\n text format where the first three columns displayed the Euler angles \u03c61, \u03a6 and \u03c62 in radians. Columns 4 and 5 display\n the coordinate Xg and Zg, respectively, for a given measurement point. It should be noted that the actual data column 5\n in the .ang file are incorrectly labeled Y. Again, the label Y in column 5 corresponds to the global direction Zg. Columns\n 7 and 8 display the image quality (IQ) and confidence index (CI) associated with the collected Kikuchi patterns.\n\n 3D Dataset - voxelized structure.\u2003 The raw Kikuchi patterns obtained from EBSD measurement are pro-\n vided in the .up2 binary file format for each EBSD slice from the TriBeam tomography measurement, these\n files can be opened using EMsoft dictionary indexing30,31, EMSphInx32 spherical pattern reindexing or with the\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 11www.nature.com/scientificdata/ www.nature.com/scientific\n\n## 4. stinville2022multimodaldatasetof pages 12-13 (Context ID: pqac-00000003)\n\n 7 and 8 display the image quality (IQ) and confidence index (CI) associated with the collected Kikuchi patterns.\n\n 3D Dataset - voxelized structure.\u2003 The raw Kikuchi patterns obtained from EBSD measurement are pro-\n vided in the .up2 binary file format for each EBSD slice from the TriBeam tomography measurement, these\n files can be opened using EMsoft dictionary indexing30,31, EMSphInx32 spherical pattern reindexing or with the\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 11www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n\n\n\n Fig. 17\u2002 Experimental validation by slip trace imaging. (a) HR-DIC strain field \u03b5xx for a reduced region of\n interest. Intense strain bands show the location of slip events. (b) The associated in-plane intensity of slip given\n in nanometers. (c) BSE image for the same region of interest. BSE imaging is an established SEM modality\n to image slip events. (d) Overlap of the HR-DIC measurement (slip intensity) with the BSE image. Perfect\n agreement is observed between BSE imaging and HR-DIC measurement.\n\n\n commercial EDAX or Oxford EBSD analysis software. The data can be downloaded from the DRYAD repos-\n itory29; however, the up2 datafiles are each 3.26 GB resulting in a large approximately 1.71 TB download. The\n reconstructed DREAM3D dataset can be downloaded independently from DRYAD29 for the interested reader\n who does not require the raw up2 EBSD patterns. The indexed data from each slice are also provided in the\n EMsoft HDF .h5 formatted files. The BSE and SEM images obtained after each slice during the TriBeam exper-\n iment are provided in 8-bit tiff images. The reconstructed 3D data with the distortion correction applied is pro-\n vided in the .dream3d file (h5 structure). The Hierarchical Data Format version 5 (h5), is an open source file\n format that supports large, complex, heterogeneous data. H5 file uses a \u201cfile directory\u201d like structure that allows\n one to organize data within the h5 file in many different structured ways, as one might do in a standard computer\n file system. The h5 format also allows for embedding of metadata making it self-describing. The associated .xdmf\n file provides the container data descriptions. Such files can be opened using the open-source software Paraview\n (https://www.paraview.org/) for visualization. For more information about the different conventions used to store\n the 3D EBSD data, we encourage the reader to consult the following references18,19.\n\n 3D Dataset - meshed structures.\u2003 The mesh generated with XtalMesh is provided in two file formats, .inp\n and .vtk. The .inp file, labelled XtalMesh.inp, is used as input into ABAQUS. The .inp file format is standard to\n ABAQUS and is an ASCII data file that consists of a series of ABAQUS keyword and data lines. In it, all nodes,\n elements, and element sets of the mesh are defined. There, the mesh is represented by ten-node tetrahedral ele-\n ments (C3D10). The .vtk file, labelled XtalMesh.vtk, is a binary Visualization Toolkit (VTK) datafile provided\n for ease of visualization and analysis of the mesh using ParaView by Kitware. There, the mesh is represented by\n four-node tetrahedral elements (C3D4) to reduce system load during visualization.\n The mesh generated with the software suite from Simmetrix (SimModeler Voxel), is provided in an Ansys33\n input deck. The input deck contains the following APDL (Ansys Parametric Design Language) files: run_Model.\n mac, contains all the calls to load the mesh, assign material properties, define node components, orient each\n grain according to the average Euler angles from the EBSD measurement and, for example, assign boundary\n conditions. The file In718RVE.cdb contains element connectivity and nodal coordinates of the entire RVE\n model, assignWPCSYS.mac creates coordinate systems using Euler angles (each coordinate system ID is linked\n to the element coordinate system defined for each grain), EulerAngles.txt contains the averaged Euler angles\n (available in the .dream3d file) for each grain in the reconstructed microstructure, In718RVE_pd.cdb assigns\n element coordinate systems (ESYS command in APDL) to each element cluster that defines a grain and creates\n node components on each side of the RVE for an easy assignment of boundary conditions, assignMatProp.mac\n assigns material properties (cubic elasticity constitutive model2) to all grains, and In718RVE_ss.cdb performs\n the finite element solution. The finite element model setup presented above was tested to make sure a solution\n can be obtained with the Ansys solver.\n\n\n\nScientific Data | (2022) 9:460 | https://doi.org/10.1038/s41597-022-01525-w 12www.nature.com/scientificdata/ www.nature.com/scientificdata\n\n\n\n\n\n Fig. 18\u2002 Experimental validation of the grain boundary locations. (a) Inverse pole figure map for a reduced\n region of interest obtained from the EBSD measurement performed at the surface of the specimen before\n TriBeam serial sectioning. (b) FeatureID maps at the surface of the 3d dataset of the same region of interest\n obtained after TriBeam serial sectioning and 3D reconstruction. (c) Overlap of (a and b). Very good agreement\n is obtained between the grain morphology from conventional EBSD measurement and TriBeam serial\n sectioning.\n\n\n\n\n\n Fig. 19\u2002 Experimental validation of the crystallographic orientation. Crystallographic orientation as displayed\n in the pole figure 001 of the grain labelled 820 in Fig. 18 from conventional EBSD measurement at the surface\n of the specimen and TriBeam serial sectioning. A minimal misorientation is observed between the two\n experimental measurements.\n\n\n Technical Validation\n HR-DIC Measurements.\u2003 Initial validation of the high-resolution strain fields obtained by HR-DIC is\n achieved by directly comparing the calculated average \u03b5xx strain along the loading direction over the representa-\n tive region of interest with the measured macroscopic strain along the gauge length of the specimen. HR-DIC and\n macroscopic strains are reported in Table 1.\n Good agreement is observed between the macroscoscopic total strain and average \u03b5xx from the HR-DIC\n measurements. The differences in percent of the total deformation between the two measurements are provided\n in brackets in Table 1 for the different load steps. Further validation of the HR-DIC measurements is provided\n by direct comparison of the HR-DIC fields to high resolution BSE images and displayed in Fig. 17 for a reduced\n region of interest. BSE imaging is a common measurement technique for capturing the location and existence\n of slip events at sample surfaces. Every single slip event is captured by the HR-DIC measurement and their loca-\n tions are identical to those identified in the BSE images. The intensity of the strain concentration and in-plane\n slip obtained by the HR-DIC measurement and using the Heaviside-DIC method was previously validated10. The\n intensity of slip is obtained for eac\n\n## 5. kaufmann2020phasemappingin pages 1-2 (Context ID: pqac-00000004)\n\n Microscopy and Microanalysis (2020), 1\u201311\n\n doi:10.1017/S1431927620001488\n\n\n Original Article\n\n Phase Mapping in EBSD Using Convolutional Neural Networks\n\n\n Kevin Kaufmann1 , Chaoyi Zhu2, Alexander S. Rosengarten1, Daniel Maryanovsky3, Haoren Wang1\n and Kenneth S. Vecchio1,2*\n 1Department of NanoEngineering, UC San Diego, La Jolla, CA 92093, USA; 2Materials Science and Engineering Program, UC San Diego, La Jolla, CA 92093, USA and\n 3Department of Cognitive Science, UC San Diego, La Jolla, CA 92093, USA\n\n\n Abstract\n\n The emergence of commercial electron backscatter diffraction (EBSD) equipment ushered in an era of information rich maps produced by\n determining the orientation of user-selected crystal structures. Since then, a technological revolution has occurred in the quality, rate detec-\n tion, and analysis of these diffractions patterns. The next revolution in EBSD is the ability to directly utilize the information rich diffraction\n patterns in a high-throughput manner. Aided by machine learning techniques, this new methodology is, as demonstrated herein, capable of\n accurately separating phases in a material by crystal symmetry, chemistry, and even lattice parameters with fewer human decisions. This\n work is the first demonstration of such capabilities and addresses many of the major challenges faced in modern EBSD. Diffraction patterns\n are collected from a variety of samples, and a convolutional neural network, a type of machine learning algorithm, is trained to autono-\n mously recognize the subtle differences in the diffraction patterns and output phase maps of the material. This study offers a path to\n machine learning coupled phase mapping as databases of EBSD patterns encompass an increasing number of the possible space groups,\n chemistry changes, and lattice parameter variations.\n\n Key words: convolutional neural network, crystal structure, EBSD, electron diffraction, machine learning\n\n (Received 26 December 2019; revised 1 April 2020; accepted 21 April 2020)\n\n\n\n Introduction systems (McLaren & Reddy, 2008; Karthikeyan et al., 2013).\n Despite its shortcomings, the Hough transform has continued\n Conventional electron backscatter diffraction (EBSD) is a scan-\n to be the standard in pattern indexing, even though computing\n ning electron microscope (SEM)-based technique used to deter-\n power has increased exponentially, largely because computer algo-\n mine the three-dimensional orientation of individual grains in\n rithms have previously been incapable of Kikuchi band detection,\n crystalline materials. At present, the collected electron backscatter\n or the more challenging problem of autonomously parsing relevant\n diffraction patterns (EBSPs) are primarily utilized to construct\n information for determining chemistry and the crystal structure.\n representative maps of the microstructure, given user-defined\n The inability for state-of-the-art Hough-based EBSD systems\n phases. The resulting maps and EBSPs can be further utilized\n to make determinations about individual components of or the\n to study dislocation evolution (Zhu et al., 2018) and geometrically\n entire crystal structure of a phase has not gone unnoticed in the\n necessary dislocations (Zhu et al., 2016), among others\n scientific community. One proposed solution is a dictionary-\n (Wilkinson et al., 2006; Schwartz et al., 2009; Engler & Randle,\n based approach (Park et al., 2013; Chen et al., 2015; Singh &\n 2010; Wilkinson & Britton, 2012). These analyses are currently\n De Graef, 2017; Ram & De Graef, 2018). Libraries of selected\n achieved in commercial EBSD utilizing a Hough transform on\n materials can be selected and simulated with software, such as\n each diffraction pattern combined with a look-up table of inter-\n EMSoft (Callahan & De Graef, 2013), for the purpose of compar-\n planar angles constructed from a set of selected reflectors as spec-\n ing the experimentally collected diffraction patterns with those\n ified by the user (typically up to five phases). In the process, the\n simulated. Through this process, the most similar patterns in\n Hough transform strips most of the information about the crystal\n the dictionary determine the phase and orientation of each\n structure and chemistry from the image (Michael & Goehner,\n EBSD pattern, even for deformed or fine-grained materials\n 1999; Dingley & Wright, 2009; Li & Han, 2015; Nolze et al.,\n (Ram & De Graef, 2018). These approaches have helped to allay\n 2017). The small subset of information being utilized for differen-\n major challenges in phase differentiation, such as separating mar-\n tiating the phases often results in mis-classification of similar\n tensite from cementite (Ram & De Graef, 2018). Other types of\n crystal structures (Chen & Thomson, 2010) and orientations\n phase identification methodologies have been demonstrated,\n that produce similar diffraction maxima in different crystal\n wherein EBSD is combined with other analytical techniques,\n such as energy-dispersive X-ray spectroscopy (EDS) or wave-\n length-dispersive X-ray spectroscopy (WDS), given that the\n *Author for correspondence: Kenneth S. Vecchio, E-mail: kvecchio@eng.ucsd.edu\n chemical and structural information of the phase exists in a the- Cite this article: Kaufmann K, Zhu C, Rosengarten AS, Maryanovsky D, Wang H,\n Vecchio KS (2020) Phase Mapping in EBSD Using Convolutional Neural Networks. oretical model or crystal database (Goehner & Michael, 1996;\n Microsc Microanal. doi:10.1017/S1431927620001488 Nowell & Wright, 2004; Dingley & Wright, 2009). The problem\n\n \u00a9 Microscopy Society of America 2020\n\n\nDownloaded from https://www.cambridge.org/core. Uppsala Universitetsbibliotek, on 12 May 2020 at 08:11:30, subject to the Cambridge Core terms of use, available at\nhttps://www.cambridge.org/core/terms. https://doi.org/10.1017/S1431927620001488 2 Kevin Kaufmann et al.\n\n\n remains that the collected EBSD images are traditionally pro- is being done in other applications (Ziletti et al., 2018; Oviedo\n cessed via the Hough transform when attempting to perform et al., 2019). In well-known classes of materials, application-\n phase identification. As mentioned previously, the down sampling specific models could be developed to distinguish phases tradi-\n of this data allows for patterns from multiple crystal structures to tional Hough-based EBSD finds nearly indistinguishable, such\n be mistaken for others since the computer algorithms are capable as martensite and ferrite. Moreover, modern computing hardware\n of finding orientations which may only appear to be correct allows for real-time classification in line with recent advance-\n (Karthikeyan et al., 2013). In commercial phase ident\n\n## 6. kaufmann2020phasemappingin pages 6-7 (Context ID: pqac-00000005)\n\nmposite. (a) Electron image showing the multiphase microstructure. (b) EDS map of Fe. (c) Hough transform EBSD map with\n user-selected libraries. (d) Phase map generated by Hough transform combined with non-optimized EDS measurements. (e) Phase map generated by Hough trans-\n form combined with optimized EDS measurements as reference chemistry. (f) Phase map generated from the CNN-based model\u2019s analysis of each diffraction pat-\n tern individually. Scale bars = 500 \u03bcm. There are 10,164 total EBSPs (pixels).\n\n\n This results in two body-centered cubic phases (space group 229), Wang et al. (2019) and their observation that crystal orientation\n pure Fe and \u03b1-Fe from the Al substitution in the lattice, and two influences the diffusion rates. Otherwise, the phase map is in\n primitive cubic phases (space group 221), nearly equiatomic FeAl good agreement with the expected results, and the CNN-based\n B2 and the non-equiatomic B2 solid solution. The backscattered EBSD method is adept at separating diffraction patterns based\n electron image in Figure 4a shows the alternating layers of the on space groups and small changes in chemistry within the\n material and gives an early indication that multiple phases are same space group.\n present in the sample. Figure 4b provides a reference Fe map CNNs applied to EBSD can resolve another challenging prob-\n and the bright columns indicate where pure Fe layers remain lem that phase identification including EDS reference chemistry\n after the reaction. If the FeAl B2 and pure Fe diffraction libraries for each phase does not completely resolve. The electron image\n are selected by the user, the Hough-based EBSD is unable to dif- for a cycled thermal barrier coating containing five FCC phases\n ferentiate between the two phases, and a phase map as shown in and one rhombohedral phase is shown in Figure 5c. The top\n Figure 4c is produced. Because the diffraction maxima are so sim- layer is yttria-stabilized zirconia (YS\u2013ZrO2), followed by the ther-\n ilar between these two phases, every diffraction pattern is identi- mally grown oxide layer (Al2O3). In the middle section is a com-\n fied as both phases with equal confidence, and the software plex bond coat containing Ni, a chrome-carbide phase (Cr23C6),\n selects, by default, the second phase on the user-selected list. TaC, and a small amount of Al2O3. The bottom layer is the nickel-\n Figure 4d demonstrates a representative result from a user assign- based superalloy Hastelloy X with Cr23C6 at the grain boundaries\n ing a reference chemistry from within each phase, using point from thermal cycling. Except for the Al2O3 phase, each of these\n EDS, to each of the four phases present. If the user selects the ref- phases belongs to space group 225, meaning that their crystal\n erence chemistry for each phase, with the best of intentions and symmetries are the same, but the lattice parameters can be differ-\n accuracy, a resultant map as shown in Figure 4d can be achieved. ent. As seen in Figure 5a, the Hough transform method alone is\n While the four phases are now each present, the phase map does unable to reliably differentiate the FCC phases and indexes them\n not accurately describe what is observed in the chemistry analysis. primarily as Hastelloy X with some interspersed YS\u2013ZrO2. The\n Instead, the pure Fe and \u03b1-Fe (Fe-SS) layers are interspersed ran- phase fraction of TaC (the brightest spots in the electron\n domly (rather than layers), and the B2-SS is scattered throughout image) is much too large (based on the known chemistry) and\n the region known to be near-equiatomic FeAl B2. After consider- is incorrectly included in the substrate. The Hough transform per-\n able \u201coptimization\u201d (repeatedly changing the reference chemistry forms reasonably well at identifying the band of Al2O3, which is\n selected in point EDS for each phase), the user can eventually not surprising given its rhombohedral structure. Including refer-\n reach a phase map (as shown in Fig. 4e) that looks similar to ence chemistry for each of the phases improves the overall quality\n the known microstructure; this of course can only be achieved of the Hough-based EBSD phase map, as shown in Figure 5b. The\n because the microstructure in this case is known in advance. band of TGO is further defined and the YS\u2013ZrO2 layer contains\n By comparison, a CNN trained on a small subset of patterns much fewer erroneous pixels. However, the Hastelloy X substrate\n belonging to each phase produces the phase map seen in has large grains and twins that are being identified as TaC and the\n Figure 4f. While the phases are not as perfectly linear as in Hastelloy X is indexed as approximately 40% of the bond coat\n Figure 4e, it is likely that the merging of phases at their borders layer (center section). Furthermore, the phase fraction of TaC in\n is real to some degree according to the analysis performed by the bond coat is still much higher than what is observed to be\n\n\nDownloaded from https://www.cambridge.org/core. Uppsala Universitetsbibliotek, on 12 May 2020 at 08:11:30, subject to the Cambridge Core terms of use, available at\nhttps://www.cambridge.org/core/terms. https://doi.org/10.1017/S1431927620001488 Microscopy and Microanalysis 7\n\n\n\n\n\n Fig. 5. Phase mapping a thermal barrier coating. (a) Hough transform EBSD map with user-selected libraries. (b) Phase map generated by Hough transform com-\n bined with optimized EDS measurements as reference chemistry. (c) Electron image showing the multiphase microstructure. The Hastelloy X substrate is at the\n bottom of the image. (d) Phase map generated from the CNN-based model\u2019s analysis of each diffraction pattern individually. (e\u2013l) EDS maps of the analyzed\n region. Scale bars = 100 \u03bcm. There are 45,052 total EBSPs (pixels).\n\n\n\n present in the electron image. Figure 5d is the result of training a reference pattern is typically unknown. This method is not com-\n CNN-based model to differentiate these phases. While not free of pared herein owing to the need for a dictionary of simulated pat-\n errors, the resulting phase map is observed to be a higher-fidelity terns for the initial calibration. The microstructure of commercial\n mapping of the area shown in the electron image than traditional 430 stainless steel is displayed in Figure 6a. While not yet obvious\n Hough-based EBSD offers. Comparison with the EDS maps in from the electron image alone, the martensite regions appear\n Figures 5e\u20135l further confirms the increased plausibility of the raised compared to the ferrite. The band contrast map in\n phase fractions and the credible identification of each diffraction Figure 6b confirms both the location of the martensite as well\n pattern. The YS\u2013ZrO2 layer is nearly completely indexed as such as the overlap in pattern quality between the martensite and fer-\n with most of the errors present residing at pores and cracks in the rite. This is the same problem encountered in the previous work\n oxide. The Al2O3 is now indexed as a continuous band of ther- (Goulden et al., 2017) and explains why their pattern quality-\n mally grown oxide in good agreement with the morphology of based model was dependent on iterative phase discrimination\n the Al EDS map. The central region is almost entirely indexed and \u201clocking\u201d pixels the user deemed correctly identified before\n as Ni with interspersed TaC, Cr23C6, and Al2O3. Careful compar- re-analysis. The region depicted in Figure 6a is mapped via tradi-\n ison with the EDS maps for Ta, Cr, Al, and O reveal the chemistry tional EBSD (Fig. 6c) using ferrite and martensite libraries gener-\n to overlap well with the presence, or absence, of the respective ated based on the lattice parameter from XRD and\n phase in the CNN-based phase map. The Hastelloy X substrate well-established equations from the literature (Xiao et al., 1995)\n is well confined and interlaced with Cr23C6 particles that were (refer to the Methods section for more information). The\n not identified in Figures 5a and 5b but are clearly present in Hough-based method results in nearly all the 12,236 collected dif-\n the Cr EDS map. fraction patterns being identified as martensite. In contrast, the\n The final demonstration of the potential for CNNs to revolu- machine learning model trained on \u201cstandards\u201d for martensite\n tionize EBSD technology is perhaps the most universally \n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005]\n\nStatus: Paper Count=25 | Relevant Papers=3 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=6 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_G0lw5LzH1ACLfx203yBWOksS", + "info": { + "start_ts": 755.583425656, + "end_ts": 759.976118886 + } + }, + { + "role": "tool", + "content": "## 1. burnett2016largevolumeserial pages 4-6 (Context ID: pqac-00000006)\n\ns conducted using a Nikon Metrology 225/320 kV\nWC-11 wt% Co sample were also prepared using an FEI Nova Na- Custom Bay system. The voltage used was 80 kV and the current\nnoLab FIB using a Ga\u00fe LMIS source at 30 kV and currents of 20 nA 130 \u03bcA. 3142 projections were acquired with an exposure time of\nand 1 nA. To provide a benchmark for comparison a Gatan Ilion 1 s per projection. The entire volume was reconstructed at full\nAr\u00fe Broad Ion Beam (BIB) Precision Polishing and Etching System resolution using a FDK reconstruction algorithm associated with\n(PECS) was also used to prepare a surface of the WC-11% Co the acquisition hardware. The resolution of the reconstructed vo-\nsample by removing 250 nm of material. A three-step polishing lume was (2 \u03bcm)3.\nprocedure in 2 10 3 Pa vacuum used a range of BIB conditions: Using measurements and orientation from the X-ray CT data a\n(i) 610 mA at 10 kV for 32 min., (ii) 450 mA at 7 kV for 10 min., and site for serial sectioning was prepared using the PFIB to capture a\n\ufb01nally (iii) 350 mA at 6 kV for 10 min. The sample was rocked particular region of interest where a large pore 0.5 mm below\ncontinuously by 7 30\u00b0 at a rate of 12 oscillations/min, with the the fracture surface could be seen (see Fig. 4a). The methodology\nsample surface exposed to BIB at glancing angle. to achieve this is described in detail by Burnett et al. [45]. Using\n For all the differently prepared surfaces of the WC-11 wt% Co 30 kV and a current of 1.3 \u03bcA the top as well as the sides and the\nsample identical EBSD acquisition parameters were used to allow front of a block were prepared ready for serial sectioning. For the\ndirect comparisons. EBSD maps were collected at 20 kV, 4 nA preparation of the top of the block 470 nA and 180 nA were used in\ncurrent and 14 mm working distance. The following EBSD condi- sequence to ensure a smooth top surface. 470 nA, 180 nA and \ufb01-\ntions were used: 50 nm step size, 6 ms dwell time, averaged nally 59 nA were used to prepare the front of the block face all at\nframes 1, 4 4 binning, 12 bands, RA and Hough\u00bc50 giving an 30 kV.\nacquisition rate of 164 Hz. A block of dimensions 135 \u03bcm across and 115 \u03bcm deep of\n 3D serial sectioning was conducted on a block of WC-11 wt%Co the A508 Grade 3 sample was serial sectioned. A total of 186 slices\nhard metal 150 120 80 mm3 in size. 790 slices were acquired were recorded with each 80 nm thick slice taking 6 min 20 s T.L. Burnett et al. / Ultramicroscopy 161 (2016) 119\u2013129 123\n\n\n\n\n\nFig. 4. (a) A virtual section from the X-ray CT data for the 0.5 mm diameter core machined from below a fracture surface caused by ductile fracture of a A508 Grade\n3 pressure vessel reactor steel compact tension specimen [32], (b) SEM image showing the same region being prepared by PFIB milling ready for serial sectioning.\n\n\nwith 3 min 25 s for the image acquisition and the remaining 2 min 4 4 binning, 6 bands, RA and Hough\u00bc60 giving an acquisition\n55 s predominantly milling time and some limited stage move rate of 95 Hz.\ntime (o10 s). Serial sectioning of this sample was conducted using\n59 nA current and 30 kV acceleration voltage. It was determined 2.3. Serial sectioning procedure\nexperimentally that a beam current of 59 nA was appropriate for\nserial section milling to obtain a surface \ufb01nish of suf\ufb01cient quality Fig. 5 shows how a region of interest is typically prepared for\nto then image the microstructure using the TLD detector. A rocking serial sectioning. The volume of interest is \ufb01rst protected by de-\nmill angle of 74\u00b0 was used (see Section 2.3). SEM images were positing a protective top layer of (a Pt layer of 100 100 mm2 and\ncollected using 2 kV, 800 pA and the through lens detector (TLD), a 1 mm thick is deposited with in about 10 minutes) covering the RoI\nhorizontal \ufb01eld width of 160 \u03bcm with a resolution of 6144 4096 and extending suf\ufb01ciently far to provide a uniform region where\ngave a pixel size of 26 nm. the tracking \ufb01ducials can also be milled. A front trench is milled to\n Both of the stacks of SEM images acquired from the serial create a cross section and is accompanied by the excavation of side\nsectioning were aligned using the align slices module using a least trenches, which not only de\ufb01ne the block for serial sectioning but\nsquares algorithm in Avizo 9.0.0. The aligned images were \ufb01ltered also control re-deposition during milling. Thus a block of material\nusing the Fourier Filter (FFT Filter) which removes periodic noise is de\ufb01ned which is to be serially sectioned. These blocks for serial\nsuch as the vertical curtaining effects produced from the rocking sectioning are typically prepared with a multistep milling proce-\nmilling process as such the FFT \ufb01lter was used at two angles that dure using the PFIB at 30 kV and several beam currents, starting\nmatch the rocking mill angles. A sharpening \ufb01lter was subse- with fast bulk cutting at 1.3 mA, followed by gentler milling at\nquently applied to compensate for some blurring resulting from 180 nA to de\ufb01ne the \ufb01nal shape. In each case the maximum cur-\nthe FFT \ufb01lter. For the A508 Grade 3 data an edge preserving \ufb01lter rent that is used to give the required surface \ufb01nish needs to be\nand a median \ufb01lter were then applied to improve the de\ufb01nition of determined for each different type of material.\nthe voids, inclusions and carbides. The data was then segmented To successively reveal a series of equally spaced and parallel\nusing a semi-automated approach which combined a watershed slices, each with the exposed block face undamaged, we have\nsegmentation routine to label the large voids and a top hat seg- applied and re\ufb01ned the rocking milling process integrated within a\nmentation routine to label the carbides, both procedures are in- fully automated routine. Once a cross-section is prepared (see\ncorporated into Avizo 9.0.0 software. Fig. 5) in the region of interest rocking milling was used, alter-\n The EBSD map obtained from the A508 Grade 3 steel was ac- nating between 7\u03b1 ( 8\u00b0r\u03b1r8\u00b0) relative to the conventional\nquired using a 20 kV electron beam energy, 4 nA current and vertical, top\u2013down, milling. This procedure is illustrated in Fig. 6.\n10 mm working distance. The following EBSD conditions were To start, the sample was in the imaging position (sample tilted to\nused: 75 nm step size, 2.3 ms dwell time, averaged frames 4, 52\u00b0 and 0\u00b0 rotation) so that the ion beam is incident to the normal\n\n\n\n\n\nFig. 5. Typical preparation of a region for serial sectioning tomography data acquisition prepared using the PFIB. A multi-step preparation procedure used the PFIB at 30 kV\nand 1.3 mA, 180 nA and 59 nA to prepare the block for the serial sectioning. With the sample tilted to 52\u00b0 and with 0\u00b0 rotation the block face is presented to the electron beam\n(a) SEM image and (b) PFIB ion beam image of the same region (a vertical top down view).124 T.L. Burnett et al. / Ultramicroscopy 161 (2016) 119\u2013129\n\n\n\n\n\nFig. 6. Rocking Polish Routine explained showing a diagrammatic view inside the FIB-SEM chamber (a) shows the sample in the imaging position of 52\u00b0 tilt and 0\u00b0 rotation\n(b) rotation of 90\u00b0 and tilt at 47\u00b0 for the rocking polish step at 5\u00b0 (c) rotation of 90\u00b0 and tilt 5\u00b0 for the rocking polish step at \u00fe5\u00b0. The magenta beam represents the electron\nbeam and the cyan the ion beam.\n\n\nof the sample surface and the block face is presented to the elec- from this position the stage moved the sample back to the imaging\ntron beam for imaging of the block face (Fig. 6a). The sample was position. This sequence (Fig. 6a) b) a) c) a) b) a) c) \u2026\u2026) was re-\nthen subjected to a 90\u00b0 compucentric rotation (i.e. rotation of the peated to build up the series of sections.\nsample by 90\u00b0 around the centre of the image). The sample is then Within the automated routine ion beam tracking \ufb01ducial\ntilted (which due to the rotation is now a tilt parallel to the cross markers (see Fig. 5) are used and recognised providing the pla-\nsection length), in this example case the sample was then tilted to cement of the milling location of the cross-sectional block face\n47\u00b0 giving a \u03b1\u00bc 5\u00b0 offset (Fig. 6b). The face was milled at this slices. To improve accuracy three different images of the \ufb01ducials\nposition before returning to the imaging position. After the image are automatically recorded, one from each of the different tilt\nof the block face was recorded the sample was rotated to 90\u00b0 and positions \u03b1\u00bc 5\u00b0, 0\u00b0 and \u00fe5\u00b0. This avoid\n\n## 2. burnett2016largevolumeserial pages 6-7 (Context ID: pqac-00000007)\n\n the image). The sample is then Within the automated routine ion beam tracking \ufb01ducial\ntilted (which due to the rotation is now a tilt parallel to the cross markers (see Fig. 5) are used and recognised providing the pla-\nsection length), in this example case the sample was then tilted to cement of the milling location of the cross-sectional block face\n47\u00b0 giving a \u03b1\u00bc 5\u00b0 offset (Fig. 6b). The face was milled at this slices. To improve accuracy three different images of the \ufb01ducials\nposition before returning to the imaging position. After the image are automatically recorded, one from each of the different tilt\nof the block face was recorded the sample was rotated to 90\u00b0 and positions \u03b1\u00bc 5\u00b0, 0\u00b0 and \u00fe5\u00b0. This avoids inaccuracies induced by\nthen tilted to 57\u00b0 giving the \u03b1\u00bc \u00fe5\u00b0 offset (Fig. 6c). After milling distortion of the \ufb01ducial image when recorded from the tilt\n\n\n\n\n\nFig. 7. EBSD results acquired under the same conditions showing representative phase maps (PM), band contrast (BC) maps and raw patterns for WC -11 wt% Co for surfaces\nprepared by (a) BIB at 6 kV and 350 mA, (b) PFIB at 30 kV and 59 nA, (c) FIB at 30 kV and 20 nA and (d) FIB at 30 kV and 1 nA. The band contrast maps are presented using the\nsame greyscale for comparison. Arrows indicate Co HCP phase. T.L. Burnett et al. / Ultramicroscopy 161 (2016) 119\u2013129 125\n\npositions. The FEI Slice and ViewTM (version 1.6) software in- with 1 nA using the Ga\u00fe FIB compared to 59 nA in the Xe\u00fe PFIB.\ncorporates a rocking polish as described above and was used in The way we have assessed these results is by considering what the\nthis work. maximum usable current is in order to achieve the results required.\n This is somewhat dif\ufb01cult to describe as there are many aspects to\n be considered when it comes to determining what is an appro-\n3. Results and discussion priate level of quality. As discussed in the introduction damage\n introduced by ion beam milling has multiple aspects to it with\n3.1. Milling damage study using a WC\u2013Co hard metal sample regards to chemical and physical interactions. As mentioned in the\n introduction the reactivity of Ga\u00fe is a clear but unquanti\ufb01ed dif-\n Given the susceptibility of WC\u2013Co hard metals to Ga\u00fe FIB ferentiator compared to the chemically inert Xe\u00fe ions. We ac-\nmilling damage [4,12], this material has been used to assess the knowledge that we discuss the cumulative effect of damage and in\nPFIB milling quality. Previously we have found that Ar\u00fe BIB pol- reality there are several convoluted aspects to this but most no-\nishing introduces only low levels of damage into this system tably surface amorphization, implantation and induced lattice\n[10,11] and so it is used as a benchmark here. The electron back- strain. A far more detailed study is needed but is beyond the scope\nscatter diffraction patterns generated from the Ar\u00fe BIB prepared of this work. The results presented here thus provide a benchmark\nsurface (Fig. 7a) are very high quality with high indexing rates where we have found that we can utilise 60 higher current\n(90%) for the WC phase and high band contrast metrics for Co (1 nA vs. 59 nA) to prepare surfaces of roughly equivalent quality\n(245/255) (the metrics of the average greyscale level are given out as determined by EBSD pattern quality from band contrast and\nof a full scale of 255 for the band contrast from 8 bit EBSD maps) indexing rate recorded under identical conditions. We are also\nand WC (180/255) (see Table 2). likely to have bene\ufb01tted from an improved sputtering rate, al-\n The corresponding results for the PFIB operated at 30 kV and though this has not been not quanti\ufb01ed.\n59 nA are shown in Fig. 7b. Vertical top\u2013down milling was used to An intriguing result is that there does appear to be an in\ufb02uence\nprepare this surface to give fair comparison with the FIB results, as of the current on the level of damage introduced to the specimen\n independent from the accelerating voltage. A current of 65 nA isa result some curtaining can be seen which could be removed by\n available in the FIB but the resulting quality of the \ufb01nished surfaceusing the rocking polishing technique described in Section 2.3. The\n that means that 1 nA must be used to prepare this surface whenEBSD indexing rate is only marginally worse, at 86% for the WC\n the FIB is used. Considerable systematic work is required to mapphase, than that recorded for the low damage BIB milling (see\n this parameter space of accelerating voltage and current for bothTable 2) with good values for WC (108/255) and excellent average\n the Ga\u00fe and Xe\u00fe milling parameters. The possibility of a chemicalband contrast for Co (220/255).\n interaction of the Ga\u00fe with the WC is a clear possibility but un- It is evident from Fig. 7c that poor results are achieved when\n quanti\ufb01ed here. Although glancing angle milling has been em-using the Ga\u00fe FIB at 20 nA (30 kV). The quality of the WC EBSD\n ployed the higher current may have led to a higher dose of Ga\u00fepatterns are degraded and are virtually unindexed (indexing only\n being imparted into the sample. Another possibility is that the8%) with very poor band contrast (40/255).\n spherical aberration which leads to the Ga\u00fe LMIS spot size rapidly Fig. 7d shows that when the FIB is operated at 1 nA and 30 kV\n increasingly 45 nA [32] means that even 20 nA is ineffective for\nthe quality of the EBSD patterns is much improved (see Table 2)\n preparing a cross section of this sensitive material for EBSD ana-\nbeing only slightly inferior to the results achieved with the PFIB\n lysis. It is possible that the greater \u2018tails\u2019 on the cutting beam and a\noperated at 59 nA current with an indexing rate in the WC phase\n divergence from the ideal grazing incidence of the ions impinging\nof 76% and a band contrast average greyscale of 94/255.\n on the cross section result in a deeper penetration and more ser-\n With regard to the Co phase, the band contrast is only slightly\n ious physical damage. Further work is needed and custom aper-\nworse than for BIB and PFIB, but there is some evidence (blue\n tures should be used to avoid limitations relating to discretization\nareas) that some regions of the Co matrix transformed from the\n of the currents that can be used i.e. the present con\ufb01guration of\ncubic into the hexagonal phase ( 3% HCP). The fact that different\n the PFIB provides currents of 59, 180, 470 and 1300 nA at the high\nareas were mapped due to the different times taken to prepare the\n end.\nsurfaces mean that we cannot make a quantitative comparison but\nit appears that the extent of transformation observed (taking into\n 3.2. Serial sectioning of WC\u2013Co sample\naccount regions that have developed during sintering [4]) in the\nFIB section are greater than in the PFIB and BIB datasets. It is\n Fig. 8 shows the ability of the PFI\n\n## 3. winiarski2021correctionofartefacts pages 15-17 (Context ID: pqac-00000008)\n\nis of each section, of WC-11%wt.Co\nhard metal data collected previously [11]. We present a new methodology for SE-EBSD image\npair co-registration based on trainable Combination of Shifted Filter Responses filters and two-\ndimensional Moving Least Squares method. Our approach and it\u2019s validation relies on three\nassumptions, 1) Each of our slices are planar, 2) SE images taken on the SEM, at 0\u00b0 stage tilt,\nare distortion free and 3) the approach validation was done on the reasonable expectation of\nthe large WC grains to have planar facets, which are reported elsewhere for similar CW-Co\nsystems [32,50,58\u201361] and are also seen in etched samples [32] and the abundance of WC\ngrains with straight grain edges or boundaries in the 2D images and EBSD analysis of polished\nsurfaces. The alignment of the corrected facets with the expected crystallographic planes was\nfurther validation of the corrected 3D-EBSD data. This allowed 3D-EBSD data to be corrected\nand used for quantitative analyses and 3D visualisation. This is applied to our hard metal case\nstudy giving rise to the following conclusions:\n\u2022 The cumulative long-range distortions, in the EBSD maps compared to the reference SE\n images, were quantified as being of the order of up to 10 \uf06dm across a 200 \u00d7 200 \uf06dm area\n section. They were found to significantly affect the shapes of individual grains in 2D\n leading to significant errors in grain morphologies in 3D. We should note that the distortion\n corrections effect the shape of the grains and not the orientation.\n\u2022 Curtaining effects were observed on the serial sections and the topography was measured\n using confocal microscopy and reported in our earlier paper [11], but the confocal\n microscopy measurements were not used in the correction methods here and is a subject of\n future work. However, due to the inclined angle of incidence lead to significant distortions\n (a few pixels, up to 0.3 \uf06dm) [11] of EBSD maps given that the grains are of the order of the\n grid used for long-range correction, these distortions required short-range correction. Here\n an attempt has been made to use to correct the distortions in a 3D-EBSD dataset of a hard\n metal case, with respect to the SE images and the expectation of the 3D shapes of the hard\n metal grains from the chemical extracted method as discussed earlier in section 3.\n\u2022 3D visualisation clearly showed that the volume of reconstructed microstructure using the\n raw data was severely deformed, where surface topography for a flat facet can be as high as\n 3.3 \uf06dm.\n\u2022 Using only the long-range correction, the topography for a planar facet, of a large well-\n defined WC grain, was reduced to 2.9 \uf06dm and the normal of the planar facet lay within about\n 10\u00b0 \u2013 20\u00b0 crystallographic positioning error.\n\u2022 By applying the short-range corrections, the distortion corrections reduced the maximum\n roughness of the planar facet surface topography to 0.4 \uf06dm and the misalignment of the\n surface of the grain facets to their respective crystallographic planes were reduced to about\n 7\u00b0 \u2013 12\u00b0.\nThese types of distortion correction algorithms are required to improve the accuracy of the\nEBSD maps. In this study the accuracy was based on the roughness of the reconstructed grainfacets, where a suitable material was chosen that the large WC grains are known to have planar\nfacets. Going forwards more automated procedures need to be developed so that large area\nEBSD maps can be routinely corrected and a more robust measure of accuracy needs to be\ndeveloped. This is especially important when reconstructing 3D data because registry from\nlayer to layer is important if features are not to be introduced into the reconstructed grain\nmorphologies. The correction methods outlined could be applied for 2D EBSD maps and 3D\nEBSD maps collected at different length-scales using other SST methods, e.g. mechanical\nsectioning, SEM microtome, FIB-SEM tomography and the more recently developed\ntechniques developed for large area milling such as BIB, plasma and laser technologies\n[15,16,68]. Furthermore, smart sampling techniques may be used for measuring accurate 3D\nshapes with only the imaging data and reduce the time required for orientation data\nmeasurement.\n\u2022 While the methods described above quantify and correct for in plane distortion, work also\n needs to be done to correct for the errors caused by uneven slice thickness during 3D\n reconstruction for visualisation to account for the actual spacing and misalignment of the\n successive layers in 3D-EBSD stacks. This effect has previously been shown to be\n significant [11]. However, use of the topography information gathered by confocal\n microscopy for correction is not trivial and requires further work to incorporate the\n topography measurements into the 3D visualisation software. This requires the software to\n be able to handle non-orthogonal layers, which is not the case at the moment and is the\n subject of future work.\n\n\n5. Acknowledgements\n\nThis work was supported by the grant from the University of Manchester \u2013 EPSRC \u2013 NPL research\nprogram: 3D Assessment of Surface Integrity and Performance (EP/H020047/1). Authors acknowledge\nGATAN for lending the prototype Illion broad ion beam system. The facilities now lie within, and are\nmaintained through, the Henry Royce Institute through EPSRC grants EP/R00661X/1, EP/S019367/1,\nEP/P025021/1 and EP/P025498/1. PJW is grateful to the ERC for advanced grant funding through\nCORREL-CT Grant No. 695638.\n\n\n\nDeclaration of interests\n\n\u2612 The authors declare that they have no known competing financial interests or personal\nrelationships that could have appeared to influence the work reported in this paper.\n\n\n\n6. References\n\n\n[1] G. Nolze, Image distortions in SEM and their influences on EBSD measurements,\n Ultramicroscopy. 107 (2007) 172\u2013183. https://doi.org/10.1016/j.ultramic.2006.07.003.\n[2] S. Maraghechi, J.P.M. Hoefnagels, R.H.J. Peerlings, M.G.D. Geers, Correction of scan line\n shift artifacts in scanning electron microscopy: An extended digital image correlation\n framework, Ultramicroscopy. 187 (2018). https://doi.org/10.1016/j.ultramic.2018.01.002.\n[3] R. Flatab\u00d8, A. Coste, M.M. Greve, A systematic investigation of the charging effect in\n scanning electron microscopy for metal nanostructures on insulating substrates, J. Microsc.\n 265 (2017). https://doi.org/10.1111/jmi.12497.\n[4] L.A. Giannuzzi, F.A. Stevie, Introduction to focused ion beams: Instrumentation, theory,\n techniques and practice, 2005. https://doi.org/10.1007/b101190.\n[5] B.E. Sorensen, J. Hjelen, H.W. \u00c5nes, T. Breivik, Recent features in EBSD, including new trapezoidal correction for multi-mapping, in: IOP Conf. Ser. Mater. Sci. Eng., 2020.\n https://doi.org/10.1088/1757-899X/891/1/012021.\n[6] A.J. Schwartz, M. Kumar, B.L. Adams, D.P. Field, Electron backscatter diffraction in\n materials science, 2009. https://doi.org/10.1007/978-0-387-88136-2.\n[7] D. Rowenhorst, Removing Imaging Distortions Through Automatic Stitching of EBSD\n Mosaics, Microsc. Microanal. 19 (2013). https://doi.org/10.1017/s1431927613006193.\n[8] A.L. Pilchak, A.R. Shiveley, J.S. Tiley, D.L. Ballard, AnyStitch: A tool for combining electron\n backscatter diffraction data sets, J. Microsc. 244 (2011). https://doi.org/10.1111/j.1365-\n 2818.2011.03496.x.\n[9] A. Winkelmann, G. Nolze, G. Cios, T. Tokarski, P. Ba\u0142a, Refined calibration model for\n improving the orientation precision of electron backscatter diffraction maps, Materials (Basel).\n 13 (2020). https://doi.org/10.3390/ma13122816.\n[10] B. Winiarski, G.S. Schajer, P.J. Withers, Surface Decoration for Improving the Accuracy of\n Displacement Measurements by Digital Image Correlation in SEM, Exp. Mech. (2012).\n https://doi.org/10.1007/s11340-011-9568-y.\n[11] B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G.E. Thompson, P.J. Withers, Broad ion beam\n serial section tomography, Ultramicroscopy. (2017).\n https://doi.org/10.1016/j.ultramic.2016.10.014.\n[12] F.J. Humphreys, Grain and subgrain characterisation by electron backscatter diffraction, J.\n Mater. Sci. (2001). https://doi.org/10.1023/A:1017973432592.\n[13] T.L. Burnett, P.J. Withers, Completing the picture through correlative characterization, Nat.\n Mater. (2019). https://doi.org/10.1038/s41563-019-0402-8.\n[14] G. Spanos, D.J. Rowenhorst, A.C. Lewis, A.B. Geltmacher, Combining serial sectioning,\n EBSD analysis, and image-based finite element modeling, MRS Bull. (2008).\n https://doi.org/10.1557/mrs2008.124.\n[15] M.P. Echlin, T.L. Burnett, A.T. Polonsky, T.M. Pollock, P.J. Withers, Serial sectioning in the\n SEM for three dimensional materials science, Curr. Opin. Solid State Mater. Sci. (2020)\n 100817. https://doi.org/10.1016/j.cossms.2020.100817.\n[16] A. Gholinia, M. Curd, E. Bousser, K. Taylor,\n\n## 4. winiarski2021correctionofartefacts pages 4-6 (Context ID: pqac-00000009)\n\nlity of the data can be compromised, such as in cases where there is\npoor pattern quality, low signal to noise and signal to background ratios, multiple phases with\nlow symmetry, pseudo-symmetry and complex crystal structures and highly deformed\nmicrostructures. This means EBSD data are often mapped at a dwell time per pixel of around\none to three orders of magnitude longer than SE imaging, which results in acquisition times of\nhundreds of minutes per map.\n\n As a result large area scans can be undertaken rapidly by SE imaging, where image\ndistortions and drift can be kept to less than a pixel [10], by using a combination of suitable\nbeam dwell time, image pixel resolution and high enough magnification to minimise lens\ndistortions, where applications can be found in digital image correlation (DIC) techniques [33\u2013\n36]. However, acquiring an EBSD map can be much slower, therefore the electron beam or\nthe sample are prone to drifts away from the assumed position leading to long-range image\ndistortions during the acquisition of an EBSD map [1,37,38]. Furthermore, SE images are\nusually taken at 0\u00b0 stage tilt or incidence angle, while EBSD maps are taken at 70\u00b0 stage tilt or\nincidence angle. Therefore, the EBSD maps have 1/cosine(70\u00b0) or 2.9 times smaller beam\nspacing or deflection in the direction perpendicular to the tilt axis in order to generate an\nequidistant grid of measurements points on the sample surface. This results in almost a\nthreefold degradation in spatial resolution and increase of any effect from drift or \u2018long-range\ndistortion\u2019, furthermore it accentuates the distortion from topographic features or \u2018short-range\ndistortion\u2019, in the EBSD maps [12]. Thus, these distortion exaggerations are minimised during\nSE imaging at 0\u00b0 incidence angle. This is also one of the advantages of the relatively new\ntechnique of transmission EBSD (t-EBSD) or transmission Kikuchi diffraction (TKD), where\nthe sample is usually back tilted only between 0\u00b0-20\u00b0 [39\u201342]. New approaches to EBSD\nanalysis, such as the \u2018Dictionary based indexing\u2019 [43,44] or smart acquisition [45] promise\nfaster acquisition times without compromising EBSD indexing rates, because either lower\nquality EBSD patterns and faster EBSD mapping speeds can be used or already utilizes other\nSEM imaging modes taken at higher speeds and requires less sampling points. Nevertheless,\nEBSD analysis will continue to be much slower than conventional image acquisition and thus\nmore susceptible to drifts.\nIn a previous paper introducing serial section broad ion beam tomography we identified\ndistortions in large area 2D EBSD maps [11]. It was shown that these can seriously affect the\ninterpretation of 3D EBSD datasets if uncorrected through an exemplar study of the 3D\nmicrostructure of WC-Co hard metals. In this follow-up paper we return to this problem to\nexamine the nature and to quantify the magnitude of the long-range and short-range distortions\nassociated with acquiring large area 2D EBSD data. All imaging modes in the SEM can have\ndistortions, however to correct for the large distortions encountered in the EBSD maps we\nassume that SE images are distortion free compared to the EBSD maps. Furthermore, we\npropose methods for correcting both of long and short-range distortions by comparing EBSD\nmaps to SE images. Our proposed methods utilise COSFIRE, Combination of Shifted Filter\nResponses filters [46] and Moving Least Squares (MLS) rigid deformation method for digital\nimage morphing [47]. The aim of this paper is to raise awareness of these artefacts which are\nlargely neglected at present, to quantify them and to show the benefits of their correction. This\npaper is not aimed at proposing the best method of correction, nor comparing the merits of\nspecific correction approaches, but rather to encourage a greater focus on developing such\nmethods going forwards.\n By way of a case study, this paper focuses on distortions associated with collecting\nrelatively large volume serial section data on WC-Co hard metals (or cermets). They are used\nfor machining, cutting, mining and drilling and are challenging because they contain a high\nfraction of a ceramic phase (i.e. the WC grains) held together a ductile metallic phase (i.e. theCo). The mechanical properties of cemented carbides strongly depend on the size and\nmorphology of the WC grains [48,49]. Abnormal grain growth (AGG) can have a detrimental\neffect on the mechanical properties of cermets, since these oversized grains (typically > 10 -\n30 \u00b5m) act as nucleation points for cracking, breakage and drill bit failure. The mechanism for\nAGG is not yet sufficiently understood. Some attempts have been made to simulate AGG in\ncemented carbide verified with experimental work and 2D analyses [50], but the plethora of\ninformation that can be extracted from 3D analyses allows the opportunity to much more fully\nstudy the AGG phenomenon. It must be noted here that there are non-destructive techniques\navailable to study AGG during in-situ heating [51,52], however the resolution of these\ntechniques are presently about 1 \uf06dm. Although FIB-SEM based techniques are destructive,\nbut are able to resolve volumes in 3D in nanometer resolution. Large area plasma FIB [25],\nBIB-SEM [16], laser-PFIB-SEM [24] or PFIB spin milling methods [27] are needed to capture\ndata over microstructurally relevant large volumes. As a result, such EBSD data provides a\ngood case for quantification and correction of long and short-range corrections described in\nthis paper.\n\n\n\n\n\n2. Experimental procedure and SE-EBSD image pair co-registration\n\n As mentioned above, the focus of the current paper is on the quantification of long and\nshort-range distortions associated with large area EBSD mapping and methods for their\ncorrection rather than on making serial sectioning experiments per se. To this end we return to\nthe 3D EBSD dataset collected by serial section BIB for the WC-Co hardmetal sample\ndescribed in [11].\n\n Sample\n\n The material studied was a Tungsten carbide and Cobalt (WC-Co) hardmetal containing\n11 mass % Co and having the average grain diameter of 5 \uf06dm [53]. The alloy was produced by\nMarshalls Hard Metals Ltd using conventional powder metallurgical methods. Tungsten\ncarbide (WC) has a hexagonal crystal structure with lattice parameters a=2.9\u00c5 and c=2.83\u00c5.\nThe stable Cobalt (Co) phase has a cubic crystal structure with a lattice parameter size a=3.56\u00c5.\n\n\n\n\n Broad ion beam sectioning\n The experimental BIB serial sectioning workflow followed in this study is described in\ndetail in a companion paper [11]. In essence, the sample was manually passed between a Gatan,\nIlion broad ion beam system and a dual beam FEI NOVA600i FIB-SEM. The milling\nconditions were 5 kV, 60 \uf06dA with 20 minutes mill time per slice, where the average slice\nthickness was 700 nm. After each material removal step a secondary electron (SE) image of\nthe surface was collected using the SEM at 0\u00b0 sample tilt, after which the crystallographic\ninformation is collected by EBSD mapping at 70\u00b0 sample tilt, to form SE-EBSD image pairs.\nThe quality of Ar+ ion polish achieved by with the Ilion BIB results in a very good surface for\nEBSD analysis without significant damage or phase transformation facilitating high indexing\nrates (above 90%). The co-registration of successive images was achieved using markers made\nby FIB [11]. SE-EBSD image pairs\n Presently, it is possible to undertake EBSD analysis for areas of several mm2\ndimensions by combining scanning of a large area by moving the electron beam and the\nstitching multiple scanned areas using the stage [6\u20138]. Although the size of the maps analysed\nhere are relatively modest compared to the largest ones reported elsewhere the distortion\ncorrections applied here are applicable to much larger areas too. The EBSD maps used in this\nwork were acquired at 200 nm step size with map size of 205 \u00d7 177 \uf06dm (1024 \u00d7 884 pixels)\nand were taken at 70\u00b0 angle of incidence and sample tilt using a Nordlys-S EBSD detector from\nOxford Instruments. This EBSD map size is relatively large for an ion based serial section\ntomography (SST) technique. The EBSD settings were chosen for the purpose of 3D-EBSD\nanalysis. Therefore, to ensure an accurate and high indexing percentage and to minimise mis-\nindexing for the WC and Co phases, a relatively high resolution EBSD detector setting of 4 \u00d7\n4 binning (336 \u00d7 256 pixels) was used, which means that the indexing time was about 25 ms\nper point, equivalent to 40 Hz, which was about 6 hours acquisition time per slice. Therefore\nthe EBSD settings used here resulted in an unusually slow speed for gathering an EBSD map\nper slice, however this meant that the average indexing rate was very high (> 96%) [11] and\nthe risk of low quality EBSD data was minimised for every slice. Furthermore, the slow EBSD\nmapping speed exaggerated any possible distortion effects from drift. Note that the Nordlys-S\nEBSD de\n\n## 5. stanek2020comparisonofsegmentation pages 1-2 (Context ID: pqac-00000010)\n\n Kovove Mater. 58 2020 301\u2013319 301\n DOI: 10.4149/km 2020 5 301\n\n\n\n Comparison of segmentation of 2D and 3D EBSD measurements\n in polycrystalline materials\n\n\n J. Stan\u02c7ek1*, J. Kope\u010dek2, P. Kr\u00e1l3, I. Kara\ufb01\u00e1tov\u00e14, F. Seitl4, V. Bene\u01614\n\n\n\n 1Department of Mathematics Education, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n 2Institute of Physics of the CAS, Department of Functional Materials, Na Slovance 2, 182 00 Prague, Czech Republic\n 3Institute of Physics of Materials, ASCR, \u02c7Zi\u017ekova 22, 616 62 Brno, Czech Republic\n 4Department of Probability and Mathematical Statistics, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n\n\n Received 30 December 2019, received in revised form 26 April 2020, accepted 17 June 2020\n\n\n Abstract\n\n A comparison study of grain segmentation in 2D and 3D electron backscatter di\ufb00rac-\n tion (EBSD) measurements of polycrystalline materials, based on a misorientation threshold,\n is presented. Using Laguerre tessellations with orientation marks, a simulation study that\n demonstrates a di\ufb00erent sensitivity of segmentation in the case of low angle and high angle\n grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investi-\n gated concerning segmentation, and there are given some recommendations for the misorien-\n tation thresholds. Computational tools such as DREAM 3D software enable signi\ufb01cant data\n evaluation that is completed by a \ufb01ve-parameter analysis of grain boundaries.\n\n\n K e y w o r d s: electron backscatter di\ufb00raction (EBSD), grain boundaries, misorientation,\n tessellation\n\n\n\n 1. Introduction observed the boundaries \u2013 the separators, whereas\n grains or rather their interiors were homogeneous ar-\n The electron backscattered di\ufb00raction method eas inside with de\ufb01ned or expected value of properties.\n(EBSD) has been a powerful tool in material science Of course, it is known that a grain structure with the\nsince the nineties, when a certain level of computeri- tessellation formula can be created [7]; nevertheless,\nzation and application of Hough transform took place, EBSD is the \ufb01nal step reversing axiom to construct\nsee [1, 2]. Nowadays, we can perform not only 2D the grains [2]. We have a set of measured points in the\nEBSD but exploiting focused ion beams (FIB), even EBSD method. We have positions of such points and\n3D EBSD, by materials\u2019 slicing tomography [3]. The their marks (e.g., Euler angles). Thus, here the inte-\n3D EBSD data evaluation is a complicated process [4]. rior is a quality, which exists, whereas the boundary\nStill, we expect to be able to construct grains [5] and has to be constructed. The easiest way how to create\nmicrostructure parameters by applying the same or grains from the set of measured points is the following\nsimilar procedures as in 2D EBSD data evaluation [6]. de\ufb01nition: the grain interior shall have the same or\nHowever, a discrepancy in the segmentation of data very similar properties. Thus, we construct grains by\npoints into the grains in the 2D and 3D cases is ap- collecting the measured points with the de\ufb01ned prop-\nparent, since a nonconvex grain in 3D can be observed erty. The metallographic and EBSD grains must be\nas more than one grain in a 2D slice. then the same in the optimal case. However, the re-\n The segmentation problem is deeply connected to sult of an experiment is not optimal. It contains some\nthe basic understanding of the polycrystals. Gener- noise or even a kind of systematic aberrations, and the\nally, the de\ufb01nition of grains as they were observed by reconstruction of grain structure can be done only up\nfounders of metallography was purely natural: They to some level of accuracy.\n\n*Corresponding author: tel.: +420 951553215; e-mail address: stanekj@karlin.m.cuni.cz302 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n The problem described above was omitted in com-\nputational materials science since the 3D data is fre-\nquently used in an elementary manner. Despite the\nexplosion of three-dimensional methods in the last\ndecades, the majority of 3D models use simple math-\nematical schemes to build the structure by a general\nprocedure approaching the reality by just a few pa-\nrameters, measured from 2D cuts of material [8].\n Data obtained by 3D EBSD or 2D EBSD form a\nset of orientations of measured points in a 3D ma-\ntrix (the points are called voxels in the sequel) or\n2D matrix (pixels in the sequel), respectively. The\nEBSD method considers a grain to be the set of all\npoints (voxels or pixels, respectively) with a similar\norientation, which are surrounded by a boundary of\na given misorientation threshold. The typical thresh-\nold in the EDAX OIM software is 5\u25e6[9], but other\nchoices are possible as well [10, 11]. The obtained seg-\nmentation into the grains depends on the chosen mis- Fig. 1. Di\ufb00erences between slices processed from 2D EBSD\n and the same slice processed from 3D EBSD. The greyorientation threshold, of course, but even if we use a\n colour displays the real grains (in the plane, which is or-\n\ufb01xed misorientation threshold, we may get di\ufb00erent\n thogonal to the cutting plane). The blue colour displays\nresults of the segmentation in 2D and 3D, as shown their 3D segmentation in the cutting plane, i.e., the parts\nin [12]. Therefore, the results obtained in 2D EBSD are considered to belong to the same grain. Finally, the\nmay not be easily compared with the results from 3D red and the orange colours show their 2D segmentation\nEBSD. in which they are considered to be di\ufb00erent grains. (a)\n In the paper, we present a comparison study of The situation with a nonconvex grain or a grain with a\nsegmentation of 2D and 3D EBSD measurements in boundary inside. (b) The situation with convex grains,\n boundaries with the misorientation smaller than the cho-polycrystalline materials, based on a misorientation\n sen threshold are denoted by the dashed line.\nthreshold. The main results of this topic are presented\nin Sections 2 and 3. The aim is to draw the reader\u2019s\nattention to the danger of obtaining misleading quan-\ntitative results about the geometry of microstructure crostructure analyses make use of DREAM 3D soft-\nwhen paying little attention to a proper segmentation. ware [14].\nFirst, in Section 2, the problem is demonstrated in a Some conclusions from Sections 2 and 3 are given\nsimulation study. A generalized Laguerre tessellation in more detail in the \ufb01nal part of the paper.\nmodeling of polycrystalline microstructure measured\nby 3D EBSD was used in [13]. We can simulate this\nmodel with a priori given distribution of misorienta- 2. Di\ufb00erences between 2D and 3D\ntions. The in\ufb02uence of this distribution on the seg- segmentation\nmentation is shown.\n In Section 3, we discuss the segmentation prob- The di\ufb00erences between segmentation into grains\nlem with two materials, namely the AlMgSc alloy af- in 2D and 3D are caused by the fact that in a 2D\nter ECAP and annealing, and pure copper after high- cutting plane, two sets of voxels may be surrounded\npressure torsion and annealing at room temperature. by a boundary de\ufb01ned by a given threshold of misori-\nThe polycrystalline microstructures are observed by entation so that the sets are detected as two grains,\n3D EBSD and 2D EBSD. The approaches presented while in 3D case, all these points belong to one larger\nin Section 2 for simulated toy data examples are dis- set which is surrounded by a boundary. Therefore in\ncussed subsequently for both materials. The curve of the 3D case, all these points lie in the same grain,\ndependence of the mean number of grains, in a single see Fig. 1. This behaviour can be observed when the\n2D slice, on the misorientation tolerance level, is plot- grains are nonconvex, or when there is a boundary of\nted for both 2D and 3D meth\n\n## 6. stanek2020comparisonofsegmentation pages 7-9 (Context ID: pqac-00000011)\n\negmentation\n with three misorientation thresholds is studied. The\n \ufb01rst one is 0.1\u25e6(called 0 % misorientation segmen-\n tation in the sequel), the second one is chosen so\n that 10 % of boundaries have lower misorientation\n (called 10 % misorientation segmentation in the se-\n quel). The last one corresponds to 15 % of boundaries\n with misorientation lower than the chosen threshold\n (called 15 % misorientation segmentation in the se-\n quel). The segmentation in 3D with such thresholds\n is shown in Fig. 8. The behaviour for all three cases\n (A), (C), and (D) is very similar. The 10 % misorien-\n tation segmentation contains larger grains, which are\n not presented in 0 % misorientation segmentation, and\n all three 15 % misorientation segmentations contain a\n single giant grain that covers most of the area of the\n slice. Note that the slice studied in the case (A) is\n the slice number 50, which has already been consid-\n ered above. Note that the results in 2D segmentation\n are similar for all three cases, i.e., the behaviour is\n similar to that one presented in Figs. 4a\u2013c. Therefore\n the segmentation in 2D is not visualized in this pa-\n per.\n Note that similar results are obtained when pro-\n cessing the other slices in the data set. Thus, based on\n this short simulation study, it can be concluded that\n the di\ufb00erence between processing of the data in 2D\n and 3D is signi\ufb01cant if at least 10\u201315 % of boundaries\n have misorientation lower than the threshold chosen\n for the segmentation procedure. The di\ufb00erence be-\n tween the processing of the data in 2D and 3D is sig-\nFig. 7. The simulated data: (a) case (C) and (b) case (D). ni\ufb01cantly a\ufb00ected by the thickness of the sample, but\n it seems that the morphology of the grains does not\n play an important part.\nated 25\u201374, and (f) segmentation in 3D with the entire\ndata set (100 slices). The result for the misorientation\nthreshold 2.5\u25e6is shown in Fig. 6. One can see that the 3. Real data\ndi\ufb00erence between the segmentation in 2D and 3D is\nmarginal when the specimen is thin (10 slices in our After the analysis of the simulated data in the pre-\ncase, i.e., 100 \u00d7 100 \u00d7 10 voxels), on the other hand, vious section, two real data sets are investigated, and\nthe di\ufb00erence is signi\ufb01cant when the specimen has 30 the segmentation into the grains in 2D and 3D is com-\nor more slices. The results are very similar if the data pared. The materials are described below in Sections\nset with 50 slices (e) or the entire data set with 100 3.1. and 3.2. It has to be added that most microstruc-\nslices (f) is used. tures show a particular orientation spread within the\n Another question is whether the grain size and grain interiors, and this is the result of the process-\nmorphology a\ufb00ect the di\ufb00erence between the segmen- ing history. It should be distinguished from acquisition\ntation in 2D and 3D as well. To investigate this de- noise, although for grain segmentation purposes, they\npendence, two more simulated data sets were added are usually not separated. Both materials have been\nto our study, namely (C) the Laguerre tessellation measured by the 3DEBSD method, which provides the\nwith a few distinctively larger grains and (D) the tes- data in the form of 2D slices. Such slices can be sub-\nsellation with nonconvex grains, see Fig. 7. The ori- sequently joined into a 3D image. Processing of the\nentation of the crystals in the cases (C) and (D) is data, such as removing noise, matching 3D slices into\nchosen in a similar way as in the case (A); there- a 3D image or removing grains, which are too small,\nfore, the misorientation between neighbouring grains etc., has been made by the software DREAM.3D. 2D\nis low, and the histograms of the misorientation are and 3D are compared in the same way as in the pre-\nsimilar to the histogram in Fig. 3a. The segmenta- vious section. First, a 2D slice processed separately308 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n\n\n\nFig. 8a\u2013f. The segmentation of the simulated data in 3D for the case (A) with the misorientation threshold (a) 0.1\u25e6, (b)\n2.39\u25e6, and (c) 2.58\u25e6. The segmentation of the simulated data in 3D for the case (C) with the misorientation threshold (d)\n 0.1\u25e6, (e) 2.22\u25e6, and (f) 2.34\u25e6.\n\n\n\nby DREAM.3D is chosen, and the obtained result is 3.1. Data of AlMgSc alloy\nquite similar to the results from working with 2D data.\nThen, the entire 3D data is processed in the same way The \ufb01rst investigated material is the aluminium\nand after 3D reconstruction, the chosen slice is cut alloy Al-3Mg-0.2Sc, which is an alloy known for its\nout. In this way, two processings of the same slice are superplasticity at high deformation rates [17]. The\nobtained (one in 2D and one in 3D), which are com- sample was subjected to severe plastic deformation\nparable. by equal channel angular pressing (ECAP), annealed J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 309\n\n\n\n\n\nFig. 8g\u2013i. The segmentation of the simulated data in 3D for the case (D) with the misorientation threshold (g) 0.1\u25e6, (h)\n 2.07\u25e6, and (i) 2.30\u25e6.\n\nsubsequently for 1 h at 400\u25e6C and investigated by\nthree-dimensional electron backscattered di\ufb00raction\n(3D-EBSD). Note that the size of the specimen was\n98.4 \u00d7 71.7 \u00d7 35.1 \u00b5m3 and it has been already investi-\ngated in detail in [12], where the di\ufb00erence between de-\ntection of grains in 2D and 3D is also brie\ufb02y discussed.\nIn the present paper, this di\ufb00erence is focused on more\ndetails. Since the sample includes a lot of noise, \ufb01lters\nin DREAM.3D have to be used to minimize the e\ufb00ect\nof the noise. The data as processed in DREAM.3D,\nwhere the grains are detected by using the \ufb01lter \u2018Seg-\nment Features (Misorientation)\u2019 with a misorientation\nthreshold of 2\u25e6, are visualized in Fig. 9. Note that the\nchoice of misorientation threshold as well as the set-\ntings for the procedure of removing noise \ufb01rmly in-\n\ufb02uence the number of detected grains. In our case, Fig. 9. The visualization of 3D data of AlMgSc alloy ex-\n cluding the noisy part of the specimen as well as very smallthe procedure detects 74 579 grains, but if we do not\n grains. The remaining part involves 74 579 grains.\nremove the noise and small grains (namely the grains\nconsisting of less than 8 voxels), the procedure detects\nmore than one million grains, where most of them are\nvery small (few voxels only). in a single slice is strongly a\ufb00ected by the noise, and\n Firstly, we compare the mean number of grains in a unfortunately, it is not clear how to remove this e\ufb00ect\nslice with the misorientation threshold chosen for the in the same way\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011]\n\nStatus: Paper Count=25 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=12 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_mH0dRIstCN2AOJxkl8ecCACI", + "info": { + "start_ts": 755.584416592, + "end_ts": 760.352480105 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_xTeD5YsxnoGXce4YcmR9STo0", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 1, \"updated_plan_objective_json\": \"{\\\"id\\\":1,\\\"objective\\\":\\\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\\\",\\\"rationale\\\":\\\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\\\",\\\"evaluation\\\":\\\"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\\\"}},{\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Error parsing plan step JSON: Failed to load JSON from text '{\"id\":1,\"objective\":\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\"rationale\":\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\"status\":\"completed\",\"result\":\"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\"evaluation\":\"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"}},{'.", + "name": "update_plan", + "tool_call_id": "call_xTeD5YsxnoGXce4YcmR9STo0", + "info": { + "start_ts": 783.853173669, + "end_ts": 783.855532371 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_Ky7P0IBBulkQ2t2Ljs7dEu3k", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 1, \"updated_plan_objective_json\": \"{\\\"id\\\":1,\\\"objective\\\":\\\"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\\\",\\\"rationale\\\":\\\"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\\\",\\\"evaluation\\\":\\\"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_Ky7P0IBBulkQ2t2Ljs7dEu3k", + "info": { + "start_ts": 805.401958358, + "end_ts": 805.403789393 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_ByYZSnq7nwSJDkOm9Cg48Zp8", + "type": "function", + "function": { + "arguments": "{\"artifact_type\": \"table\", \"description\": \"Create a concise prior-art evidence matrix with rows for the five requested areas. Columns: Area, closest published work(s), what is automated, objective/criterion, what is explicitly not automated, and gap assessment. Use only evidence already gathered. Include Polonsky 2019 grain segmentation parameter sweep; Shen 2019 CNN-confidence hybrid dictionary proposal; Zhu 2022 global optimization as a nearby per-pattern refinement work noting available metadata if evidence is limited; Sparks 2024 automated registration/pattern-center workflow; Burnett 2016 automated serial sectioning but manually/experimentally selected acquisition settings; Winiarski 2021 automated distortion correction but not indexing tuning; Kaufmann 2020 CNN phase mapping. Clearly distinguish demonstrated methods from proposals and from absent closed loops.\"}", + "name": "create_artifact" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "### Artifact (ID artifact-00)\n\n| Area | Closest published work(s) | What is automated | Objective/criterion | What is explicitly not automated | Gap assessment |\n|---|---|---|---|---|---|\n| 1. Optimizer-driven tuning of EBSD pattern-processing or indexing parameters | **Closest demonstrated EBSD parameter sweep:** Polonsky et al. 2019 parameterized DREAM.3D reconstruction variables and swept grain-segmentation misorientation tolerance over 0.1\u201310\u00b0 in 0.1\u00b0 increments, generating 100 reconstructions; **nearby but different:** Zhu et al. 2022 is only available here as metadata for \u201cglobal optimization algorithms\u201d applied to orientation/pattern-center/deformation extraction, so it should be treated as adjacent prior art, not verified from text in this record; **nearby non-tuning automation:** Kaufmann et al. 2020 CNN phase mapping automates phase recognition from patterns rather than tuning EBSD processing settings (pqac-00000014, pqac-00000013, pqac-00000004, pqac-00000005) | Automated batch execution / parameterization of reconstruction pipelines in BisQue+DREAM.3D; CNN-based direct phase classification from patterns in Kaufmann 2020 (pqac-00000014, pqac-00000004) | Polonsky used reconstruction-level metrics: fraction of \u201cgood\u201d CI-masked voxels retained and total grain count, selecting an elbow/balance point at 2.0\u00b0; Kaufmann targets phase-map fidelity qualitatively against known microstructure/EDS, not optimizer search (pqac-00000013, pqac-00000005) | No evidence here of Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, RL, random/grid search over **pattern** processing settings such as background correction, binning, neighbor averaging, Hough band count, rho-theta smoothing, or dictionary settings. Polonsky optimizes segmentation/reconstruction variables after indexing, not indexing settings themselves. Kaufmann does not tune EBSD parameters at all (pqac-00000013, pqac-00000005) | **Gap largely open.** Closest evidence is reconstruction-pipeline parameterization, not closed-loop optimizer-driven tuning of pattern-processing/indexing knobs. |\n| 2. Feedback loops that re-index stored raw patterns using CI/IQ/MAD/NCC to choose next parameter set | **Closest proposal:** Shen et al. 2019 proposes combining fast CNN predictions with a confidence measure (quaternion norm) so that high-confidence points get neighborhood dictionary search and low-confidence points get full dictionary search; **related raw-pattern availability, not loop:** Stinville et al. 2022 notes raw .up2 patterns can be opened in EMsoft DI / EMSphInx, enabling reindexing in principle (pqac-00000017, pqac-00000015, pqac-00000002, pqac-00000003) | Shen automates confidence estimation for each location and proposes routing subsequent dictionary matching effort based on confidence; raw pattern archival in Stinville enables offline reindexing workflows (pqac-00000017, pqac-00000003) | Quaternion norm as confidence proxy in Shen; no demonstrated CI/IQ/MAD/NCC-driven next-parameter selection loop in the gathered evidence (pqac-00000017) | Shen does **not** demonstrate an executed closed loop that iteratively changes indexing/processing parameters and re-indexes stored raw patterns to maximize CI/IQ/MAD/NCC/NCC-like objectives; it is a proposed hybrid search allocation strategy. Stinville provides data access, not autonomous optimization (pqac-00000017, pqac-00000003) | **Gap open.** Proposal-level hybrid confidence routing exists, but not a demonstrated feedback optimizer that selects the next EBSD parameter set from reindexed raw patterns. |\n| 3. Optimization against grain-level or microstructure-level outcomes | **Closest demonstrated:** Polonsky et al. 2019 explicitly optimizes a grain-segmentation parameter against microstructure-level outcomes (grain count and retained good-data fraction); Stan\u011bk et al. 2020 compares 2D/3D segmentation sensitivity to misorientation threshold and shows strong threshold dependence, but not autonomous optimization (pqac-00000013, pqac-00000012, pqac-00000010, pqac-00000011) | Automated parallel reconstructions and post-processing counts in Polonsky; analytical comparison of threshold effects on segmentation in Stan\u011bk (pqac-00000013, pqac-00000010) | Grain count stability / avoidance of grain merging plus retained CI-masked voxel fraction in Polonsky; sensitivity of number of grains and segmentation discrepancies vs threshold in Stan\u011bk (pqac-00000013, pqac-00000011) | No demonstrated multi-objective or constrained optimization over broader grain-level cleanup outcomes such as speckle/wild-grain fraction, grain-boundary integrity, fraction of pixels altered by cleanup, or Pareto balancing across multiple EBSD post-processing stages. Polonsky is a one-parameter sweep with manually interpreted elbow criterion (pqac-00000013, pqac-00000012) | **Partially occupied, but narrow.** Grain-level optimization exists for segmentation tolerance; broader autonomous multi-objective EBSD post-processing remains a gap. |\n| 4. Hybrid Hough-first / dictionary-on-residue and automated escalation criteria | **Closest published work:** Shen et al. 2019 explicitly proposes a hybrid strategy: CNN preliminary prediction plus confidence, then pattern matching only in a neighborhood for high-confidence points and full dictionary search for low-confidence points; this is the clearest automated escalation criterion in the gathered evidence. Sparks et al. 2024 and Winiarski et al. 2021 automate registration/distortion correction workflows but not Hough\u2192dictionary escalation (pqac-00000017, pqac-00000000, pqac-00000008) | Automated confidence scoring and proposed routing of dictionary-search scope in Shen; automated image registration/distortion correction in Sparks/Winiarski (pqac-00000017, pqac-00000000, pqac-00000008) | Criterion in Shen is confidence from quaternion norm: high confidence \u2192 local neighborhood dictionary search; low confidence \u2192 full dictionary search (pqac-00000017) | No evidence in the gathered text of a **demonstrated** Hough-first then dictionary-on-low-quality-residue loop with automated escalation based on CI/IQ/MAD/NCC thresholds. Shen\u2019s routing is proposed around CNN+dictionary, not Hough+dictionary, and not shown as an executed autonomous residue-reindexing pipeline (pqac-00000017) | **Mostly gap.** There is a close proposal for automated escalation, but not a demonstrated Hough-residue-to-dictionary autonomous workflow. |\n| 5. Autonomous/self-driving EBSD post-processing or characterization workflows, including links back to SEM acquisition recommendations | **Closest workflow automation:** Sparks et al. 2024 automates sample/image registration steps, pattern-center fitting for every section, and alignment routines; Winiarski et al. 2021 automates SE\u2013EBSD distortion correction and argues more automated procedures are needed; Burnett et al. 2016 demonstrates a fully automated serial-sectioning routine with tracking fiducials, but acquisition settings were experimentally chosen; Kaufmann et al. 2020 automates phase mapping from patterns (pqac-00000001, pqac-00000000, pqac-00000008, pqac-00000009, pqac-00000006, pqac-00000007, pqac-00000004) | Automated registration/warping, per-layer EBSD-to-BSE alignment, CI-threshold masks, pattern-center fitting per section, and automated serial sectioning with fiducial tracking; CNN inference for phase maps (pqac-00000000, pqac-00000001, pqac-00000006, pqac-00000004) | Workflow goals are geometric consistency, low inter-section misorientation, and corrected distortions; Burnett assesses equivalent quality via EBSD pattern quality/band contrast and indexing rate after experimentally selected settings (pqac-00000001, pqac-00000008, pqac-00000007) | No evidence here of a human-out-of-the-loop EBSD post-processing system that autonomously tunes post-processing/indexing parameters end-to-end, nor of a loop that feeds post-processing outcomes back to SEM acquisition parameters such as beam current, dwell time, or step size. Burnett explicitly determines acceptable milling current experimentally; Winiarski says more automated procedures still need to be developed (pqac-00000007, pqac-00000008) | **Gap open.** Sub-workflow automation exists, but not a self-driving closed-loop EBSD characterization system that recommends or controls acquisition parameters from post-processing outcomes. |\n\n\n*Table: This table maps the closest published prior art found for the five requested EBSD automation areas, distinguishing demonstrated methods from proposals and from absent closed loops. It highlights that most evidence clusters around workflow automation and grain-segmentation parameter sweeps, not autonomous closed-loop tuning of indexing/processing parameters.*\n\nArtifact IDs that may be injected into the answer: [artifact-00]\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013, pqac-00000014, pqac-00000015, pqac-00000016, pqac-00000017]\n\nStatus: Paper Count=25 | Relevant Papers=8 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=18 | Current Cost=$0.0922 | Disease-Target Evidence=0", + "name": "create_artifact", + "tool_call_id": "call_ByYZSnq7nwSJDkOm9Cg48Zp8", + "info": { + "start_ts": 829.394246022, + "end_ts": 852.626361711 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_sOxhMHmNJ6cGdwzj14gJbWOY", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 2, \"updated_plan_objective_json\": \"{\\\"id\\\":2,\\\"objective\\\":\\\"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\\\",\\\"rationale\\\":\\\"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\\\",\\\"evaluation\\\":\\\"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_Dgfm5HSnCqYdsd2pIw7qCf6a", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 3, \"updated_plan_objective_json\": \"{\\\"id\\\":3,\\\"objective\\\":\\\"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\\\",\\\"rationale\\\":\\\"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\\\",\\\"evaluation\\\":\\\"The evidence provides exact variables, criteria, and limitations.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_ERLgQRfcrgfEuXOF5b3WqhbB", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 4, \"updated_plan_objective_json\": \"{\\\"id\\\":4,\\\"objective\\\":\\\"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\\\",\\\"rationale\\\":\\\"Area 5 includes both software claims and closed-loop acquisition control.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\\\",\\\"evaluation\\\":\\\"Full text explicitly documents automation boundaries and calls for future automation.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_PqUIwWt0TdvA1AJFRXJAtK32", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 5, \"updated_plan_objective_json\": \"{\\\"id\\\":5,\\\"objective\\\":\\\"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\\\",\\\"rationale\\\":\\\"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\\\",\\\"evaluation\\\":\\\"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\\\"}\"}", + "name": "update_plan" + } + }, + { + "id": "call_Mv315OulvDSqQgHazJenPV4E", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 6, \"updated_plan_objective_json\": \"{\\\"id\\\":6,\\\"objective\\\":\\\"Create a structured artifact summarizing the evidence and limitations.\\\",\\\"rationale\\\":\\\"A compact evidence matrix will make the systematic survey auditable and useful.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Created artifact-00, a five-area evidence matrix with closest works, automation, objectives, exclusions, and gap assessments.\\\",\\\"evaluation\\\":\\\"Artifact covers all requested areas and clearly separates demonstrations, proposals, and absences.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_sOxhMHmNJ6cGdwzj14gJbWOY", + "info": { + "start_ts": 884.219772521, + "end_ts": 884.221180953 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_Dgfm5HSnCqYdsd2pIw7qCf6a", + "info": { + "start_ts": 884.221220607, + "end_ts": 884.221572424 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_ERLgQRfcrgfEuXOF5b3WqhbB", + "info": { + "start_ts": 884.221632058, + "end_ts": 884.221962904 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"completed\",\n \"result\": \"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\",\n \"evaluation\": \"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_PqUIwWt0TdvA1AJFRXJAtK32", + "info": { + "start_ts": 884.221985251, + "end_ts": 884.222258695 + } + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"completed\",\n \"result\": \"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\",\n \"evaluation\": \"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a five-area evidence matrix with closest works, automation, objectives, exclusions, and gap assessments.\",\n \"evaluation\": \"Artifact covers all requested areas and clearly separates demonstrations, proposals, and absences.\"\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_Mv315OulvDSqQgHazJenPV4E", + "info": { + "start_ts": 884.22227648, + "end_ts": 884.222507565 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_JvmVAiIMF8mMq3af5Jp56p6p", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 7, \"updated_plan_objective_json\": \"{\\\"id\\\":7,\\\"objective\\\":\\\"Submit the final technical answer with evidence-matched citations.\\\",\\\"rationale\\\":\\\"Required final delivery step.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"completed\",\n \"result\": \"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\",\n \"evaluation\": \"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a five-area evidence matrix with closest works, automation, objectives, exclusions, and gap assessments.\",\n \"evaluation\": \"Artifact covers all requested areas and clearly separates demonstrations, proposals, and absences.\"\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_JvmVAiIMF8mMq3af5Jp56p6p", + "info": { + "start_ts": 904.606249096, + "end_ts": 904.607998944 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_q8dUzD63NUAJ7NfyX8ac9nmd", + "type": "function", + "function": { + "arguments": "{\"answer\": \"# Systematic prior-art survey: closed-loop optimization of EBSD processing and indexing\\n\\n## Scope and search interpretation\\n\\nThe survey distinguished four often-confounded categories: **(i)** optimization of pattern-processing or indexing parameters, **(ii)** algorithmic refinement of orientation or pattern-center estimates for a fixed pattern, **(iii)** automated post-indexing map cleanup/segmentation, and **(iv)** automation of acquisition or registration. Only the first category squarely satisfies area 1; automation in the other categories is relevant as neighboring prior art but does not establish a closed-loop EBSD indexing optimizer.\\n\\nSearches covered EBSD with Bayesian optimization, Gaussian processes, evolutionary/genetic algorithms, particle swarm optimization, grid/random search, reinforcement learning, Hough parameters, dictionary indexing, reindexing, CI/IQ/MAD/fit/NCC, grain-level objectives, autonomous microscopy, and acquisition feedback. Patent searches returned no relevant EBSD patent in the searched corpus. Absence findings below therefore mean **\\u201cnot found in this systematic search,\\u201d not a legal opinion of novelty or freedom to operate**.\\n\\n\\n\\n## 1. Optimizer-driven tuning of pattern-processing or indexing parameters\\n\\n### Closest demonstrated work: Polonsky et al. (2019)\\n\\nPolonsky and colleagues integrated DREAM.3D into the cloud-based BisQue platform so that pipeline variables could be parameterized and executions parallelized. Their principal EBSD example varied the **post-indexing grain-segmentation misorientation tolerance** from 0.1\\u00b0 to 10\\u00b0 in 0.1\\u00b0 increments, producing 100 three-dimensional reconstructions while holding minimum-neighbor and minimum-defect-size settings fixed. They evaluated the fraction of CI-masked \\u201cgood\\u201d voxels retained and total reconstructed grain count, then selected 2.0\\u00b0 from convergence/elbow behavior in the derivatives of those metrics. This is genuine automated parameter sweeping with microstructure-level evaluation. However, it operates on already indexed orientation data; the study states that raw EBSD patterns were not collected for that dataset. It therefore does **not** tune background correction, detector binning, neighbor averaging, Hough band count, Hough rho/theta smoothing, or dictionary-search settings, and it cannot re-index patterns under candidate settings. (pqac-00000014, pqac-00000013, pqac-00000012)\\n\\n### Adjacent work that should not be overread\\n\\nShen et al. developed CNN-based orientation indexing and optimized neural-network architecture/training for speed and prediction accuracy. The trained model indexes patterns directly and uses quaternion-norm deviation as a reliability measure. This replaces or supplements conventional indexing; it does not run Bayesian, evolutionary, grid, or reinforcement-learning search over conventional EBSD processing parameters. It also remains sensitive to experimental geometry and requires detector/setup parameters to be determined beforehand or accommodated through retraining. (pqac-00000015, pqac-00000016, pqac-00000017)\\n\\nKaufmann et al. automated phase classification from individual EBSD patterns with a CNN. The work reduces human decisions in phase mapping and outperforms conventional Hough-plus-user-selected phase/chemistry workflows in difficult examples. The \\u201coptimization\\u201d described for the conventional comparator was repeated **user adjustment** of EDS reference chemistry against a microstructure already known in advance, not an algorithmic search over EBSD settings. The CNN likewise classifies patterns rather than optimizing background, Hough, binning, or dictionary parameters. (pqac-00000004, pqac-00000005)\\n\\nGlobal/evolutionary optimization has also been published for extracting or refining orientation, pattern center, and deformation state from pattern matching. That is important neighboring art: it establishes optimizer use **inside a per-pattern inverse problem**. It is not equivalent to an outer loop choosing data-processing/indexing hyperparameters from map-quality or grain-quality outcomes. The retrieved full-text evidence did not support treating such work as prior art for the broader closed-loop parameter-selection architecture.\\n\\n**Assessment:** No verified publication was found that applies Bayesian optimization, a GP surrogate, genetic/evolutionary search, random/grid search, RL, or a comparable optimizer to the listed **pattern-processing and indexing knobs** and automatically selects the best configuration. Area 1 remains substantially open, aside from adjacent per-pattern refinement and post-indexing segmentation sweeps.\\n\\n## 2. Re-indexing stored raw patterns in a metric-driven feedback loop\\n\\nStinville et al. provide a useful enabling precedent: their archived 3D dataset includes approximately 1.71 TB of raw `.up2` Kikuchi patterns, and the paper explicitly states that these can be opened and reprocessed with EMsoft dictionary indexing, EMSphInx spherical-pattern reindexing, or commercial EBSD software. The indexed files also expose IQ and CI. This demonstrates the technical availability of stored-pattern reindexing, but the paper is a dataset publication, not an optimizer: it does not generate candidate parameter sets, score them, or select the next experiment. (pqac-00000002, pqac-00000003)\\n\\nShen et al. provide the closest feedback-like decision rule. They propose obtaining a preliminary CNN orientation and confidence at every point, performing only a local dictionary search around high-confidence predictions, and performing a full dictionary search for low-confidence predictions. The confidence proxy is the norm of the predicted quaternion. This is automated computational routing, but it is not a demonstrated iterative loop over processing/indexing parameter sets, and it does not use CI, IQ, MAD/fit, or NCC to propose the next candidate configuration. (pqac-00000017)\\n\\nPolonsky et al. use CI to define an initial mask of good voxels and assess how many remain after cleanup and segmentation. CI is therefore part of a reconstruction-level quality assessment, but it is not fed back to re-index raw patterns; indeed, raw patterns were unavailable in that example. (pqac-00000013)\\n\\n**Assessment:** No demonstrated publication was found with the complete loop **candidate parameters \\u2192 re-index stored raw patterns \\u2192 compute CI/IQ/MAD/fit/NCC objective \\u2192 choose the next parameter set \\u2192 repeat**. Area 2 remains an unclaimed gap in the searched literature.\\n\\n## 3. Grain- or microstructure-level objectives\\n\\nThis area has the strongest direct prior art, but it is narrow.\\n\\nPolonsky et al. explicitly recognized that segmentation tolerance changes grain consolidation and retained data in opposite directions. Their two high-level metrics were total grain count and the fraction of initially CI-qualified voxels retained through reconstruction. They chose the lowest tolerance at which the changes in both quantities became approximately constant, analogous to an elbow criterion. This is a microstructure-level, effectively two-metric parameter-selection procedure rather than optimization of a per-pattern score. (pqac-00000013, pqac-00000012)\\n\\nStan\\u011bk et al. independently show why such objectives matter. They analyzed the dependence of 2D and 3D grain segmentation on misorientation threshold, including mean grain counts and the formation of erroneously merged giant grains. For one noisy experimental dataset, changing noise removal and excluding grains below eight voxels changed the detected population from more than one million grains to 74,579. This is a systematic sensitivity study and supplies possible grain-level observables, but it does not implement an autonomous optimizer or select cleanup parameters through a formal multi-objective rule. (pqac-00000010, pqac-00000011)\\n\\nNeither publication demonstrates optimization across a broad EBSD-processing chain using objectives such as wild-grain/speckle fraction, boundary continuity or integrity, grain-count stability over threshold perturbations, fraction of pixels modified by cleanup, or a constraint on topology preservation. Polonsky varies essentially one segmentation parameter, keeps other settings fixed, and identifies a balance/elbow rather than computing a Pareto front or solving a constrained optimization problem. (pqac-00000013)\\n\\n**Assessment:** Area 3 is **partially occupied**. A broad claim to \\u201coptimizing EBSD processing using grain-level outcomes\\u201d would encounter Polonsky et al. A narrower architecture that jointly tunes indexing and cleanup parameters using several topology- and intervention-aware grain objectives still appears open in the searched literature.\\n\\n## 4. Hybrid Hough-first/dictionary-on-residue indexing\\n\\nThe closest automated escalation rule is Shen et al., but its first stage is a CNN rather than Hough indexing. Their proposed rule is explicit: high-confidence CNN results receive dictionary matching only near the preliminary orientation; low-confidence results receive full dictionary search. Thus, the publication anticipates confidence-triggered allocation of expensive pattern matching. It does not demonstrate a Hough-first pipeline, does not define residue using CI/IQ/MAD/NCC from Hough indexing, and does not report an executed end-to-end selective reindexing experiment. (pqac-00000017)\\n\\nThe same paper compares CNN and Hough indexing at different exposure times. CNN indexing produced a larger fraction of points below a 10\\u00b0 misorientation-error cutoff at high noise, whereas Hough was more accurate among the smaller subset it indexed correctly. This comparison motivates selective hybridization, but the reported hybrid suggestion remains CNN-to-dictionary rather than Hough-to-dictionary. (pqac-00000016, pqac-00000017)\\n\\nKaufmann et al. likewise replace conventional Hough phase assignment with CNN classification; they do not retain Hough as a first pass and selectively escalate failed pixels. (pqac-00000004, pqac-00000005)\\n\\n**Assessment:** Confidence-triggered hybrid search is neighboring prior art, so a very broad claim to \\u201cuse confidence to decide whether to run dictionary matching\\u201d is not open. However, no verified demonstration was found of **Hough first \\u2192 automatically identify low-quality residue \\u2192 dictionary/pattern-match only that residue**, particularly using an explicit CI/IQ/MAD/NCC or grain-context escalation criterion. Area 4 therefore remains mostly open in the requested Hough-specific form.\\n\\n## 5. Autonomous EBSD workflows and acquisition feedback\\n\\nSeveral publications automate substantial sub-workflows:\\n\\n* **BisQue\\u2013DREAM.3D:** cloud execution, provenance, parallel pipeline runs, and range parameterization for 3D reconstruction. Parameters and ranges are supplied rather than autonomously discovered, and the demonstrated optimization begins after indexing. (pqac-00000014, pqac-00000013)\\n* **Automated serial sectioning:** Burnett et al. describe a fully automated PFIB serial-section routine using automatically recognized tracking fiducials and repeated milling/imaging positions. Yet acceptable milling current was determined experimentally, and the authors state that systematic mapping of voltage/current space remained necessary. EBSD settings\\u2014including beam energy/current, dwell, binning, band count, and Hough settings\\u2014were fixed by the investigators rather than adapted from post-processing results. (pqac-00000006, pqac-00000007)\\n* **Automated registration and distortion correction:** Sparks et al. use automated sample alignment, pattern-center fitting for each section, CI thresholding, and optimizer-based multimodal registration. The optimizer maximizes an image-registration similarity metric and transforms maps; it does not alter EBSD indexing settings or SEM acquisition conditions. (pqac-00000001, pqac-00000000)\\n* **SE\\u2013EBSD map correction:** Winiarski et al. automate aspects of co-registration and long-/short-range distortion correction using trainable filters and moving least squares. The paper explicitly concludes that more automated procedures and more robust accuracy measures still need development. It also discusses faster or smart acquisition as a prospect, not as a closed post-processing-to-acquisition controller. (pqac-00000009, pqac-00000008)\\n* **CNN indexing/phase mapping:** Shen and Kaufmann automate inference from patterns, but training data, instrument calibration, phase standards, and/or model selection remain external; neither closes the loop to microscope settings. (pqac-00000015, pqac-00000016, pqac-00000004)\\n\\nNo retrieved work used post-processing outcomes to recommend or automatically change **beam current, pattern dwell/exposure, detector binning, or spatial step size** for a subsequent SEM/EBSD acquisition. Burnett et al. come closest to connecting acquisition conditions with indexing rate and band contrast, but the comparison is an investigator-designed experiment, not a controller: selected ion currents are evaluated under fixed EBSD conditions and an acceptable current is chosen experimentally. (pqac-00000006, pqac-00000007)\\n\\n**Assessment:** Automated components and scripted pipelines are well established, so \\u201cautomated EBSD workflow\\u201d alone is not a gap. A genuinely autonomous, human-out-of-the-loop EBSD characterization system that interprets indexed/map/grain quality and then recommends or controls SEM acquisition parameters remains open in the searched literature.\\n\\n## Consolidated gap assessment\\n\\n1. **Optimizer-driven tuning of EBSD pattern-processing/indexing settings \\u2014 open.** Neighboring art covers per-pattern inverse optimization, ML indexing, and post-indexing segmentation sweeps, but not outer-loop optimization of background, binning, averaging, Hough, or dictionary hyperparameters.\\n2. **Metric-driven iterative reindexing of stored raw patterns \\u2014 open.** Raw-pattern reindexing is enabled and confidence-directed search is proposed, but the complete iterative objective loop was not found.\\n3. **Grain-level optimization \\u2014 partially claimed.** Polonsky et al. directly cover segmentation-tolerance selection using retained-good-data fraction and grain count. Broad multi-objective, constrained, topology-aware optimization across indexing plus cleanup remains open.\\n4. **Automated hybrid escalation \\u2014 partially anticipated, Hough-specific implementation open.** Shen et al. anticipate confidence-triggered local versus full dictionary search after CNN indexing. A demonstrated Hough-first, dictionary-on-residue workflow with formal escalation criteria was not found.\\n5. **Autonomous workflow with feedback to acquisition \\u2014 open at the system level.** Serial sectioning, registration, reconstruction, correction, and ML inference are individually automated, but no end-to-end post-processing-to-SEM acquisition feedback system was found.\\n\\nThe most defensible unexplored combination is therefore: **repeated reindexing of archived raw patterns under optimizer-selected processing/indexing settings; joint pattern-, pixel-, and grain-level objectives with explicit cleanup/topology constraints; confidence- and grain-context-based escalation from Hough to dictionary matching; and translation of the optimized result into acquisition recommendations or microscope control.** None of the verified publications combines those elements into a closed, autonomous EBSD loop.\"}", + "name": "answer" + } + } + ], + "info": null + } + ], + "hide_old_env_states": false, + "hide_old_action_content": false, + "sliding_window": null + }, + "observation": [ + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define scope and search systematically across scholarly literature and patents for automated EBSD parameter optimization, feedback re-indexing, grain-level objectives, hybrid indexing, and autonomous acquisition links.\",\n \"rationale\": \"The request is a systematic prior-art survey spanning five distinct technical areas and potentially patent prior art.\",\n \"status\": \"completed\",\n \"result\": \"Searched broad and exact-phrase scholarly queries across all five areas plus patent queries; retrieved 25 papers, deeply inspected the most relevant works, and found no relevant patent result in the tool corpus.\",\n \"evaluation\": \"Search covered named optimizer classes, EBSD quality metrics, Hough/dictionary hybrid routing, grain segmentation sweeps, autonomous workflows, and acquisition controls.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather and verify full-text evidence for automated/optimizer-driven tuning methods and closed-loop re-indexing objectives.\",\n \"rationale\": \"Areas 1 and 2 require distinguishing true algorithmic parameter search from ordinary fixed pipelines or cleanup.\",\n \"status\": \"completed\",\n \"result\": \"Verified no demonstrated optimizer loop over the named EBSD pattern-processing/indexing parameters in retrieved full text. Closest works are Polonsky's post-indexing segmentation sweep and Shen's confidence-routed CNN/dictionary proposal; raw-pattern archives permit reindexing but do not optimize it.\",\n \"evaluation\": \"Evidence explicitly distinguishes indexing settings from reconstruction settings and demonstrated loops from proposals.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Gather and verify evidence for grain/microstructure-level optimization and hybrid Hough-to-dictionary escalation criteria.\",\n \"rationale\": \"Areas 3 and 4 focus on outcome level and automated routing logic, which must be evidenced explicitly.\",\n \"status\": \"completed\",\n \"result\": \"Polonsky 2019 demonstrated a 100-point misorientation-tolerance sweep using retained good-data fraction and grain count; Stanek 2020 established threshold sensitivity. Shen 2019 proposed confidence-based CNN-to-dictionary escalation, but no demonstrated Hough-first/dictionary-residue workflow was found.\",\n \"evaluation\": \"The evidence provides exact variables, criteria, and limitations.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Gather and verify evidence for autonomous EBSD software/workflows and links from post-processing to SEM acquisition recommendations.\",\n \"rationale\": \"Area 5 includes both software claims and closed-loop acquisition control.\",\n \"status\": \"completed\",\n \"result\": \"Found automated sub-workflows in BisQue/DREAM.3D, serial sectioning, registration/distortion correction, pattern-center fitting, and CNN indexing/phase mapping. No post-processing-to-SEM beam-current/dwell/step-size feedback controller was found; acquisition settings were experimentally or manually selected.\",\n \"evaluation\": \"Full text explicitly documents automation boundaries and calls for future automation.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Synthesize a claim-chart-style map of closest prior art, explicitly stating what each work automates and does not automate, then assess which areas remain gaps.\",\n \"rationale\": \"The user requests a comparative map and clear gap conclusion, not merely a bibliography.\",\n \"status\": \"completed\",\n \"result\": \"Synthesized five-area claim map: areas 1, 2, and 5 remain open in searched literature; area 3 is narrowly occupied for grain-segmentation tolerance but broad multi-objective processing is open; area 4 has a close CNN-confidence proposal but Hough-to-DI demonstration remains open.\",\n \"evaluation\": \"Each conclusion is qualified as search-based rather than a legal novelty opinion and tied to verified evidence.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Create a structured artifact summarizing the evidence and limitations.\",\n \"rationale\": \"A compact evidence matrix will make the systematic survey auditable and useful.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a five-area evidence matrix with closest works, automation, objectives, exclusions, and gap assessments.\",\n \"evaluation\": \"Artifact covers all requested areas and clearly separates demonstrations, proposals, and absences.\"\n },\n {\n \"id\": 7,\n \"objective\": \"Submit the final technical answer with evidence-matched citations.\",\n \"rationale\": \"Required final delivery step.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_JvmVAiIMF8mMq3af5Jp56p6p", + "info": { + "start_ts": 904.606249096, + "end_ts": 904.607998944 + } + } + ], + "next_observation": [ + { + "role": "tool", + "content": "Answer was successfully submitted. Plan is not complete. Terminating the run anyway.", + "name": "answer", + "tool_call_id": "call_q8dUzD63NUAJ7NfyX8ac9nmd", + "info": { + "start_ts": 988.21857678, + "end_ts": 988.644864257 + } + } + ], + "reward": 0.0, + "truncated": false, + "done": true, + "value": 0.0, + "metadata": { + "time_elapsed_before_transition": 1.2038695009999856, + "time_elapsed_agent_get_asv": 65.84219959799998, + "time_elapsed_after_agent_get_asv": 1.1231598589999976, + "time_elapsed_env_step": 2.575555707000035, + "time_elapsed_after_env_step": 1.0653715199999851 + }, + "action": { + "call_id": { + "run_id": "1d44903c-5aa9-4c3d-849b-11437f062a39", + "fwd_id": "2852f4e5-73ef-4bac-89c1-b740bde8b067" + }, + "op_name": "IdentityOp-201bf506564a", + "op_class_name": "ldp.graph.common_ops.IdentityOp", + "value": { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_q8dUzD63NUAJ7NfyX8ac9nmd", + "type": "function", + "function": { + "arguments": "{\"answer\": \"# Systematic prior-art survey: closed-loop optimization of EBSD processing and indexing\\n\\n## Scope and search interpretation\\n\\nThe survey distinguished four often-confounded categories: **(i)** optimization of pattern-processing or indexing parameters, **(ii)** algorithmic refinement of orientation or pattern-center estimates for a fixed pattern, **(iii)** automated post-indexing map cleanup/segmentation, and **(iv)** automation of acquisition or registration. Only the first category squarely satisfies area 1; automation in the other categories is relevant as neighboring prior art but does not establish a closed-loop EBSD indexing optimizer.\\n\\nSearches covered EBSD with Bayesian optimization, Gaussian processes, evolutionary/genetic algorithms, particle swarm optimization, grid/random search, reinforcement learning, Hough parameters, dictionary indexing, reindexing, CI/IQ/MAD/fit/NCC, grain-level objectives, autonomous microscopy, and acquisition feedback. Patent searches returned no relevant EBSD patent in the searched corpus. Absence findings below therefore mean **\\u201cnot found in this systematic search,\\u201d not a legal opinion of novelty or freedom to operate**.\\n\\n\\n\\n## 1. Optimizer-driven tuning of pattern-processing or indexing parameters\\n\\n### Closest demonstrated work: Polonsky et al. (2019)\\n\\nPolonsky and colleagues integrated DREAM.3D into the cloud-based BisQue platform so that pipeline variables could be parameterized and executions parallelized. Their principal EBSD example varied the **post-indexing grain-segmentation misorientation tolerance** from 0.1\\u00b0 to 10\\u00b0 in 0.1\\u00b0 increments, producing 100 three-dimensional reconstructions while holding minimum-neighbor and minimum-defect-size settings fixed. They evaluated the fraction of CI-masked \\u201cgood\\u201d voxels retained and total reconstructed grain count, then selected 2.0\\u00b0 from convergence/elbow behavior in the derivatives of those metrics. This is genuine automated parameter sweeping with microstructure-level evaluation. However, it operates on already indexed orientation data; the study states that raw EBSD patterns were not collected for that dataset. It therefore does **not** tune background correction, detector binning, neighbor averaging, Hough band count, Hough rho/theta smoothing, or dictionary-search settings, and it cannot re-index patterns under candidate settings. (pqac-00000014, pqac-00000013, pqac-00000012)\\n\\n### Adjacent work that should not be overread\\n\\nShen et al. developed CNN-based orientation indexing and optimized neural-network architecture/training for speed and prediction accuracy. The trained model indexes patterns directly and uses quaternion-norm deviation as a reliability measure. This replaces or supplements conventional indexing; it does not run Bayesian, evolutionary, grid, or reinforcement-learning search over conventional EBSD processing parameters. It also remains sensitive to experimental geometry and requires detector/setup parameters to be determined beforehand or accommodated through retraining. (pqac-00000015, pqac-00000016, pqac-00000017)\\n\\nKaufmann et al. automated phase classification from individual EBSD patterns with a CNN. The work reduces human decisions in phase mapping and outperforms conventional Hough-plus-user-selected phase/chemistry workflows in difficult examples. The \\u201coptimization\\u201d described for the conventional comparator was repeated **user adjustment** of EDS reference chemistry against a microstructure already known in advance, not an algorithmic search over EBSD settings. The CNN likewise classifies patterns rather than optimizing background, Hough, binning, or dictionary parameters. (pqac-00000004, pqac-00000005)\\n\\nGlobal/evolutionary optimization has also been published for extracting or refining orientation, pattern center, and deformation state from pattern matching. That is important neighboring art: it establishes optimizer use **inside a per-pattern inverse problem**. It is not equivalent to an outer loop choosing data-processing/indexing hyperparameters from map-quality or grain-quality outcomes. The retrieved full-text evidence did not support treating such work as prior art for the broader closed-loop parameter-selection architecture.\\n\\n**Assessment:** No verified publication was found that applies Bayesian optimization, a GP surrogate, genetic/evolutionary search, random/grid search, RL, or a comparable optimizer to the listed **pattern-processing and indexing knobs** and automatically selects the best configuration. Area 1 remains substantially open, aside from adjacent per-pattern refinement and post-indexing segmentation sweeps.\\n\\n## 2. Re-indexing stored raw patterns in a metric-driven feedback loop\\n\\nStinville et al. provide a useful enabling precedent: their archived 3D dataset includes approximately 1.71 TB of raw `.up2` Kikuchi patterns, and the paper explicitly states that these can be opened and reprocessed with EMsoft dictionary indexing, EMSphInx spherical-pattern reindexing, or commercial EBSD software. The indexed files also expose IQ and CI. This demonstrates the technical availability of stored-pattern reindexing, but the paper is a dataset publication, not an optimizer: it does not generate candidate parameter sets, score them, or select the next experiment. (pqac-00000002, pqac-00000003)\\n\\nShen et al. provide the closest feedback-like decision rule. They propose obtaining a preliminary CNN orientation and confidence at every point, performing only a local dictionary search around high-confidence predictions, and performing a full dictionary search for low-confidence predictions. The confidence proxy is the norm of the predicted quaternion. This is automated computational routing, but it is not a demonstrated iterative loop over processing/indexing parameter sets, and it does not use CI, IQ, MAD/fit, or NCC to propose the next candidate configuration. (pqac-00000017)\\n\\nPolonsky et al. use CI to define an initial mask of good voxels and assess how many remain after cleanup and segmentation. CI is therefore part of a reconstruction-level quality assessment, but it is not fed back to re-index raw patterns; indeed, raw patterns were unavailable in that example. (pqac-00000013)\\n\\n**Assessment:** No demonstrated publication was found with the complete loop **candidate parameters \\u2192 re-index stored raw patterns \\u2192 compute CI/IQ/MAD/fit/NCC objective \\u2192 choose the next parameter set \\u2192 repeat**. Area 2 remains an unclaimed gap in the searched literature.\\n\\n## 3. Grain- or microstructure-level objectives\\n\\nThis area has the strongest direct prior art, but it is narrow.\\n\\nPolonsky et al. explicitly recognized that segmentation tolerance changes grain consolidation and retained data in opposite directions. Their two high-level metrics were total grain count and the fraction of initially CI-qualified voxels retained through reconstruction. They chose the lowest tolerance at which the changes in both quantities became approximately constant, analogous to an elbow criterion. This is a microstructure-level, effectively two-metric parameter-selection procedure rather than optimization of a per-pattern score. (pqac-00000013, pqac-00000012)\\n\\nStan\\u011bk et al. independently show why such objectives matter. They analyzed the dependence of 2D and 3D grain segmentation on misorientation threshold, including mean grain counts and the formation of erroneously merged giant grains. For one noisy experimental dataset, changing noise removal and excluding grains below eight voxels changed the detected population from more than one million grains to 74,579. This is a systematic sensitivity study and supplies possible grain-level observables, but it does not implement an autonomous optimizer or select cleanup parameters through a formal multi-objective rule. (pqac-00000010, pqac-00000011)\\n\\nNeither publication demonstrates optimization across a broad EBSD-processing chain using objectives such as wild-grain/speckle fraction, boundary continuity or integrity, grain-count stability over threshold perturbations, fraction of pixels modified by cleanup, or a constraint on topology preservation. Polonsky varies essentially one segmentation parameter, keeps other settings fixed, and identifies a balance/elbow rather than computing a Pareto front or solving a constrained optimization problem. (pqac-00000013)\\n\\n**Assessment:** Area 3 is **partially occupied**. A broad claim to \\u201coptimizing EBSD processing using grain-level outcomes\\u201d would encounter Polonsky et al. A narrower architecture that jointly tunes indexing and cleanup parameters using several topology- and intervention-aware grain objectives still appears open in the searched literature.\\n\\n## 4. Hybrid Hough-first/dictionary-on-residue indexing\\n\\nThe closest automated escalation rule is Shen et al., but its first stage is a CNN rather than Hough indexing. Their proposed rule is explicit: high-confidence CNN results receive dictionary matching only near the preliminary orientation; low-confidence results receive full dictionary search. Thus, the publication anticipates confidence-triggered allocation of expensive pattern matching. It does not demonstrate a Hough-first pipeline, does not define residue using CI/IQ/MAD/NCC from Hough indexing, and does not report an executed end-to-end selective reindexing experiment. (pqac-00000017)\\n\\nThe same paper compares CNN and Hough indexing at different exposure times. CNN indexing produced a larger fraction of points below a 10\\u00b0 misorientation-error cutoff at high noise, whereas Hough was more accurate among the smaller subset it indexed correctly. This comparison motivates selective hybridization, but the reported hybrid suggestion remains CNN-to-dictionary rather than Hough-to-dictionary. (pqac-00000016, pqac-00000017)\\n\\nKaufmann et al. likewise replace conventional Hough phase assignment with CNN classification; they do not retain Hough as a first pass and selectively escalate failed pixels. (pqac-00000004, pqac-00000005)\\n\\n**Assessment:** Confidence-triggered hybrid search is neighboring prior art, so a very broad claim to \\u201cuse confidence to decide whether to run dictionary matching\\u201d is not open. However, no verified demonstration was found of **Hough first \\u2192 automatically identify low-quality residue \\u2192 dictionary/pattern-match only that residue**, particularly using an explicit CI/IQ/MAD/NCC or grain-context escalation criterion. Area 4 therefore remains mostly open in the requested Hough-specific form.\\n\\n## 5. Autonomous EBSD workflows and acquisition feedback\\n\\nSeveral publications automate substantial sub-workflows:\\n\\n* **BisQue\\u2013DREAM.3D:** cloud execution, provenance, parallel pipeline runs, and range parameterization for 3D reconstruction. Parameters and ranges are supplied rather than autonomously discovered, and the demonstrated optimization begins after indexing. (pqac-00000014, pqac-00000013)\\n* **Automated serial sectioning:** Burnett et al. describe a fully automated PFIB serial-section routine using automatically recognized tracking fiducials and repeated milling/imaging positions. Yet acceptable milling current was determined experimentally, and the authors state that systematic mapping of voltage/current space remained necessary. EBSD settings\\u2014including beam energy/current, dwell, binning, band count, and Hough settings\\u2014were fixed by the investigators rather than adapted from post-processing results. (pqac-00000006, pqac-00000007)\\n* **Automated registration and distortion correction:** Sparks et al. use automated sample alignment, pattern-center fitting for each section, CI thresholding, and optimizer-based multimodal registration. The optimizer maximizes an image-registration similarity metric and transforms maps; it does not alter EBSD indexing settings or SEM acquisition conditions. (pqac-00000001, pqac-00000000)\\n* **SE\\u2013EBSD map correction:** Winiarski et al. automate aspects of co-registration and long-/short-range distortion correction using trainable filters and moving least squares. The paper explicitly concludes that more automated procedures and more robust accuracy measures still need development. It also discusses faster or smart acquisition as a prospect, not as a closed post-processing-to-acquisition controller. (pqac-00000009, pqac-00000008)\\n* **CNN indexing/phase mapping:** Shen and Kaufmann automate inference from patterns, but training data, instrument calibration, phase standards, and/or model selection remain external; neither closes the loop to microscope settings. (pqac-00000015, pqac-00000016, pqac-00000004)\\n\\nNo retrieved work used post-processing outcomes to recommend or automatically change **beam current, pattern dwell/exposure, detector binning, or spatial step size** for a subsequent SEM/EBSD acquisition. Burnett et al. come closest to connecting acquisition conditions with indexing rate and band contrast, but the comparison is an investigator-designed experiment, not a controller: selected ion currents are evaluated under fixed EBSD conditions and an acceptable current is chosen experimentally. (pqac-00000006, pqac-00000007)\\n\\n**Assessment:** Automated components and scripted pipelines are well established, so \\u201cautomated EBSD workflow\\u201d alone is not a gap. A genuinely autonomous, human-out-of-the-loop EBSD characterization system that interprets indexed/map/grain quality and then recommends or controls SEM acquisition parameters remains open in the searched literature.\\n\\n## Consolidated gap assessment\\n\\n1. **Optimizer-driven tuning of EBSD pattern-processing/indexing settings \\u2014 open.** Neighboring art covers per-pattern inverse optimization, ML indexing, and post-indexing segmentation sweeps, but not outer-loop optimization of background, binning, averaging, Hough, or dictionary hyperparameters.\\n2. **Metric-driven iterative reindexing of stored raw patterns \\u2014 open.** Raw-pattern reindexing is enabled and confidence-directed search is proposed, but the complete iterative objective loop was not found.\\n3. **Grain-level optimization \\u2014 partially claimed.** Polonsky et al. directly cover segmentation-tolerance selection using retained-good-data fraction and grain count. Broad multi-objective, constrained, topology-aware optimization across indexing plus cleanup remains open.\\n4. **Automated hybrid escalation \\u2014 partially anticipated, Hough-specific implementation open.** Shen et al. anticipate confidence-triggered local versus full dictionary search after CNN indexing. A demonstrated Hough-first, dictionary-on-residue workflow with formal escalation criteria was not found.\\n5. **Autonomous workflow with feedback to acquisition \\u2014 open at the system level.** Serial sectioning, registration, reconstruction, correction, and ML inference are individually automated, but no end-to-end post-processing-to-SEM acquisition feedback system was found.\\n\\nThe most defensible unexplored combination is therefore: **repeated reindexing of archived raw patterns under optimizer-selected processing/indexing settings; joint pattern-, pixel-, and grain-level objectives with explicit cleanup/topology constraints; confidence- and grain-context-based escalation from Hough to dictionary matching; and translation of the optimized result into acquisition recommendations or microscope control.** None of the verified publications combines those elements into a closed, autonomous EBSD loop.\"}", + "name": "answer" + } + } + ] + }, + "logprob": null + } + } + }, + "created_at": "2026-08-11T22:10:37.534862Z" + } + ], + "environment_frame": { + "id": "91591475-4fd7-4528-b6a6-f822bca8a892", + "agent_state_point_in_time": null, + "trajectory_timestep": 12, + "current_agent_step": "after_transition", + "trajectory_id": "2a641486-1ec5-49ad-b3e8-b2836932d961", + "state": { + "deepcopy": true, + "state": { + "id": "2a641486-1ec5-49ad-b3e8-b2836932d961", + "created_at": "2026-08-11T22:11:04.147640", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "request": {}, + "response": { + "answer": { + "id": "2a641486-1ec5-49ad-b3e8-b2836932d961", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "answer": "# Systematic prior-art survey: closed-loop optimization of EBSD processing and indexing\n\n## Scope and search interpretation\n\nThe survey distinguished four often-confounded categories: **(i)** optimization of pattern-processing or indexing parameters, **(ii)** algorithmic refinement of orientation or pattern-center estimates for a fixed pattern, **(iii)** automated post-indexing map cleanup/segmentation, and **(iv)** automation of acquisition or registration. Only the first category squarely satisfies area 1; automation in the other categories is relevant as neighboring prior art but does not establish a closed-loop EBSD indexing optimizer.\n\nSearches covered EBSD with Bayesian optimization, Gaussian processes, evolutionary/genetic algorithms, particle swarm optimization, grid/random search, reinforcement learning, Hough parameters, dictionary indexing, reindexing, CI/IQ/MAD/fit/NCC, grain-level objectives, autonomous microscopy, and acquisition feedback. Patent searches returned no relevant EBSD patent in the searched corpus. Absence findings below therefore mean **\u201cnot found in this systematic search,\u201d not a legal opinion of novelty or freedom to operate**.\n\n| Area | Closest published work(s) | What is automated | Objective/criterion | What is explicitly not automated | Gap assessment |\n|---|---|---|---|---|---|\n| 1. Optimizer-driven tuning of EBSD pattern-processing or indexing parameters | **Closest demonstrated EBSD parameter sweep:** Polonsky et al. 2019 parameterized DREAM.3D reconstruction variables and swept grain-segmentation misorientation tolerance over 0.1\u201310\u00b0 in 0.1\u00b0 increments, generating 100 reconstructions; **nearby but different:** Zhu et al. 2022 is only available here as metadata for \u201cglobal optimization algorithms\u201d applied to orientation/pattern-center/deformation extraction, so it should be treated as adjacent prior art, not verified from text in this record; **nearby non-tuning automation:** Kaufmann et al. 2020 CNN phase mapping automates phase recognition from patterns rather than tuning EBSD processing settings (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7) | Automated batch execution / parameterization of reconstruction pipelines in BisQue+DREAM.3D; CNN-based direct phase classification from patterns in Kaufmann 2020 (polonsky2019threedimensionalanalysisand pages 1-2, kaufmann2020phasemappingin pages 1-2) | Polonsky used reconstruction-level metrics: fraction of \u201cgood\u201d CI-masked voxels retained and total grain count, selecting an elbow/balance point at 2.0\u00b0; Kaufmann targets phase-map fidelity qualitatively against known microstructure/EDS, not optimizer search (polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 6-7) | No evidence here of Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, RL, random/grid search over **pattern** processing settings such as background correction, binning, neighbor averaging, Hough band count, rho-theta smoothing, or dictionary settings. Polonsky optimizes segmentation/reconstruction variables after indexing, not indexing settings themselves. Kaufmann does not tune EBSD parameters at all (polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 6-7) | **Gap largely open.** Closest evidence is reconstruction-pipeline parameterization, not closed-loop optimizer-driven tuning of pattern-processing/indexing knobs. |\n| 2. Feedback loops that re-index stored raw patterns using CI/IQ/MAD/NCC to choose next parameter set | **Closest proposal:** Shen et al. 2019 proposes combining fast CNN predictions with a confidence measure (quaternion norm) so that high-confidence points get neighborhood dictionary search and low-confidence points get full dictionary search; **related raw-pattern availability, not loop:** Stinville et al. 2022 notes raw .up2 patterns can be opened in EMsoft DI / EMSphInx, enabling reindexing in principle (shen2019convolutionalneuralnetworkbased pages 10-12, shen2019convolutionalneuralnetworkbased pages 1-2, stinville2022multimodaldatasetof pages 9-12, stinville2022multimodaldatasetof pages 12-13) | Shen automates confidence estimation for each location and proposes routing subsequent dictionary matching effort based on confidence; raw pattern archival in Stinville enables offline reindexing workflows (shen2019convolutionalneuralnetworkbased pages 10-12, stinville2022multimodaldatasetof pages 12-13) | Quaternion norm as confidence proxy in Shen; no demonstrated CI/IQ/MAD/NCC-driven next-parameter selection loop in the gathered evidence (shen2019convolutionalneuralnetworkbased pages 10-12) | Shen does **not** demonstrate an executed closed loop that iteratively changes indexing/processing parameters and re-indexes stored raw patterns to maximize CI/IQ/MAD/NCC/NCC-like objectives; it is a proposed hybrid search allocation strategy. Stinville provides data access, not autonomous optimization (shen2019convolutionalneuralnetworkbased pages 10-12, stinville2022multimodaldatasetof pages 12-13) | **Gap open.** Proposal-level hybrid confidence routing exists, but not a demonstrated feedback optimizer that selects the next EBSD parameter set from reindexed raw patterns. |\n| 3. Optimization against grain-level or microstructure-level outcomes | **Closest demonstrated:** Polonsky et al. 2019 explicitly optimizes a grain-segmentation parameter against microstructure-level outcomes (grain count and retained good-data fraction); Stan\u011bk et al. 2020 compares 2D/3D segmentation sensitivity to misorientation threshold and shows strong threshold dependence, but not autonomous optimization (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8, stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 7-9) | Automated parallel reconstructions and post-processing counts in Polonsky; analytical comparison of threshold effects on segmentation in Stan\u011bk (polonsky2019threedimensionalanalysisand pages 5-6, stanek2020comparisonofsegmentation pages 1-2) | Grain count stability / avoidance of grain merging plus retained CI-masked voxel fraction in Polonsky; sensitivity of number of grains and segmentation discrepancies vs threshold in Stan\u011bk (polonsky2019threedimensionalanalysisand pages 5-6, stanek2020comparisonofsegmentation pages 7-9) | No demonstrated multi-objective or constrained optimization over broader grain-level cleanup outcomes such as speckle/wild-grain fraction, grain-boundary integrity, fraction of pixels altered by cleanup, or Pareto balancing across multiple EBSD post-processing stages. Polonsky is a one-parameter sweep with manually interpreted elbow criterion (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8) | **Partially occupied, but narrow.** Grain-level optimization exists for segmentation tolerance; broader autonomous multi-objective EBSD post-processing remains a gap. |\n| 4. Hybrid Hough-first / dictionary-on-residue and automated escalation criteria | **Closest published work:** Shen et al. 2019 explicitly proposes a hybrid strategy: CNN preliminary prediction plus confidence, then pattern matching only in a neighborhood for high-confidence points and full dictionary search for low-confidence points; this is the clearest automated escalation criterion in the gathered evidence. Sparks et al. 2024 and Winiarski et al. 2021 automate registration/distortion correction workflows but not Hough\u2192dictionary escalation (shen2019convolutionalneuralnetworkbased pages 10-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17) | Automated confidence scoring and proposed routing of dictionary-search scope in Shen; automated image registration/distortion correction in Sparks/Winiarski (shen2019convolutionalneuralnetworkbased pages 10-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17) | Criterion in Shen is confidence from quaternion norm: high confidence \u2192 local neighborhood dictionary search; low confidence \u2192 full dictionary search (shen2019convolutionalneuralnetworkbased pages 10-12) | No evidence in the gathered text of a **demonstrated** Hough-first then dictionary-on-low-quality-residue loop with automated escalation based on CI/IQ/MAD/NCC thresholds. Shen\u2019s routing is proposed around CNN+dictionary, not Hough+dictionary, and not shown as an executed autonomous residue-reindexing pipeline (shen2019convolutionalneuralnetworkbased pages 10-12) | **Mostly gap.** There is a close proposal for automated escalation, but not a demonstrated Hough-residue-to-dictionary autonomous workflow. |\n| 5. Autonomous/self-driving EBSD post-processing or characterization workflows, including links back to SEM acquisition recommendations | **Closest workflow automation:** Sparks et al. 2024 automates sample/image registration steps, pattern-center fitting for every section, and alignment routines; Winiarski et al. 2021 automates SE\u2013EBSD distortion correction and argues more automated procedures are needed; Burnett et al. 2016 demonstrates a fully automated serial-sectioning routine with tracking fiducials, but acquisition settings were experimentally chosen; Kaufmann et al. 2020 automates phase mapping from patterns (sparks20243dreconstructionof pages 9-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17, winiarski2021correctionofartefacts pages 4-6, burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7, kaufmann2020phasemappingin pages 1-2) | Automated registration/warping, per-layer EBSD-to-BSE alignment, CI-threshold masks, pattern-center fitting per section, and automated serial sectioning with fiducial tracking; CNN inference for phase maps (sparks20243dreconstructionof pages 23-24, sparks20243dreconstructionof pages 9-12, burnett2016largevolumeserial pages 4-6, kaufmann2020phasemappingin pages 1-2) | Workflow goals are geometric consistency, low inter-section misorientation, and corrected distortions; Burnett assesses equivalent quality via EBSD pattern quality/band contrast and indexing rate after experimentally selected settings (sparks20243dreconstructionof pages 9-12, winiarski2021correctionofartefacts pages 15-17, burnett2016largevolumeserial pages 6-7) | No evidence here of a human-out-of-the-loop EBSD post-processing system that autonomously tunes post-processing/indexing parameters end-to-end, nor of a loop that feeds post-processing outcomes back to SEM acquisition parameters such as beam current, dwell time, or step size. Burnett explicitly determines acceptable milling current experimentally; Winiarski says more automated procedures still need to be developed (burnett2016largevolumeserial pages 6-7, winiarski2021correctionofartefacts pages 15-17) | **Gap open.** Sub-workflow automation exists, but not a self-driving closed-loop EBSD characterization system that recommends or controls acquisition parameters from post-processing outcomes. |\n\n\n*Table: This table maps the closest published prior art found for the five requested EBSD automation areas, distinguishing demonstrated methods from proposals and from absent closed loops. It highlights that most evidence clusters around workflow automation and grain-segmentation parameter sweeps, not autonomous closed-loop tuning of indexing/processing parameters.*\n\n## 1. Optimizer-driven tuning of pattern-processing or indexing parameters\n\n### Closest demonstrated work: Polonsky et al. (2019)\n\nPolonsky and colleagues integrated DREAM.3D into the cloud-based BisQue platform so that pipeline variables could be parameterized and executions parallelized. Their principal EBSD example varied the **post-indexing grain-segmentation misorientation tolerance** from 0.1\u00b0 to 10\u00b0 in 0.1\u00b0 increments, producing 100 three-dimensional reconstructions while holding minimum-neighbor and minimum-defect-size settings fixed. They evaluated the fraction of CI-masked \u201cgood\u201d voxels retained and total reconstructed grain count, then selected 2.0\u00b0 from convergence/elbow behavior in the derivatives of those metrics. This is genuine automated parameter sweeping with microstructure-level evaluation. However, it operates on already indexed orientation data; the study states that raw EBSD patterns were not collected for that dataset. It therefore does **not** tune background correction, detector binning, neighbor averaging, Hough band count, Hough rho/theta smoothing, or dictionary-search settings, and it cannot re-index patterns under candidate settings. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8)\n\n### Adjacent work that should not be overread\n\nShen et al. developed CNN-based orientation indexing and optimized neural-network architecture/training for speed and prediction accuracy. The trained model indexes patterns directly and uses quaternion-norm deviation as a reliability measure. This replaces or supplements conventional indexing; it does not run Bayesian, evolutionary, grid, or reinforcement-learning search over conventional EBSD processing parameters. It also remains sensitive to experimental geometry and requires detector/setup parameters to be determined beforehand or accommodated through retraining. (shen2019convolutionalneuralnetworkbased pages 1-2, shen2019convolutionalneuralnetworkbased pages 8-10, shen2019convolutionalneuralnetworkbased pages 10-12)\n\nKaufmann et al. automated phase classification from individual EBSD patterns with a CNN. The work reduces human decisions in phase mapping and outperforms conventional Hough-plus-user-selected phase/chemistry workflows in difficult examples. The \u201coptimization\u201d described for the conventional comparator was repeated **user adjustment** of EDS reference chemistry against a microstructure already known in advance, not an algorithmic search over EBSD settings. The CNN likewise classifies patterns rather than optimizing background, Hough, binning, or dictionary parameters. (kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7)\n\nGlobal/evolutionary optimization has also been published for extracting or refining orientation, pattern center, and deformation state from pattern matching. That is important neighboring art: it establishes optimizer use **inside a per-pattern inverse problem**. It is not equivalent to an outer loop choosing data-processing/indexing hyperparameters from map-quality or grain-quality outcomes. The retrieved full-text evidence did not support treating such work as prior art for the broader closed-loop parameter-selection architecture.\n\n**Assessment:** No verified publication was found that applies Bayesian optimization, a GP surrogate, genetic/evolutionary search, random/grid search, RL, or a comparable optimizer to the listed **pattern-processing and indexing knobs** and automatically selects the best configuration. Area 1 remains substantially open, aside from adjacent per-pattern refinement and post-indexing segmentation sweeps.\n\n## 2. Re-indexing stored raw patterns in a metric-driven feedback loop\n\nStinville et al. provide a useful enabling precedent: their archived 3D dataset includes approximately 1.71 TB of raw `.up2` Kikuchi patterns, and the paper explicitly states that these can be opened and reprocessed with EMsoft dictionary indexing, EMSphInx spherical-pattern reindexing, or commercial EBSD software. The indexed files also expose IQ and CI. This demonstrates the technical availability of stored-pattern reindexing, but the paper is a dataset publication, not an optimizer: it does not generate candidate parameter sets, score them, or select the next experiment. (stinville2022multimodaldatasetof pages 9-12, stinville2022multimodaldatasetof pages 12-13)\n\nShen et al. provide the closest feedback-like decision rule. They propose obtaining a preliminary CNN orientation and confidence at every point, performing only a local dictionary search around high-confidence predictions, and performing a full dictionary search for low-confidence predictions. The confidence proxy is the norm of the predicted quaternion. This is automated computational routing, but it is not a demonstrated iterative loop over processing/indexing parameter sets, and it does not use CI, IQ, MAD/fit, or NCC to propose the next candidate configuration. (shen2019convolutionalneuralnetworkbased pages 10-12)\n\nPolonsky et al. use CI to define an initial mask of good voxels and assess how many remain after cleanup and segmentation. CI is therefore part of a reconstruction-level quality assessment, but it is not fed back to re-index raw patterns; indeed, raw patterns were unavailable in that example. (polonsky2019threedimensionalanalysisand pages 5-6)\n\n**Assessment:** No demonstrated publication was found with the complete loop **candidate parameters \u2192 re-index stored raw patterns \u2192 compute CI/IQ/MAD/fit/NCC objective \u2192 choose the next parameter set \u2192 repeat**. Area 2 remains an unclaimed gap in the searched literature.\n\n## 3. Grain- or microstructure-level objectives\n\nThis area has the strongest direct prior art, but it is narrow.\n\nPolonsky et al. explicitly recognized that segmentation tolerance changes grain consolidation and retained data in opposite directions. Their two high-level metrics were total grain count and the fraction of initially CI-qualified voxels retained through reconstruction. They chose the lowest tolerance at which the changes in both quantities became approximately constant, analogous to an elbow criterion. This is a microstructure-level, effectively two-metric parameter-selection procedure rather than optimization of a per-pattern score. (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8)\n\nStan\u011bk et al. independently show why such objectives matter. They analyzed the dependence of 2D and 3D grain segmentation on misorientation threshold, including mean grain counts and the formation of erroneously merged giant grains. For one noisy experimental dataset, changing noise removal and excluding grains below eight voxels changed the detected population from more than one million grains to 74,579. This is a systematic sensitivity study and supplies possible grain-level observables, but it does not implement an autonomous optimizer or select cleanup parameters through a formal multi-objective rule. (stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 7-9)\n\nNeither publication demonstrates optimization across a broad EBSD-processing chain using objectives such as wild-grain/speckle fraction, boundary continuity or integrity, grain-count stability over threshold perturbations, fraction of pixels modified by cleanup, or a constraint on topology preservation. Polonsky varies essentially one segmentation parameter, keeps other settings fixed, and identifies a balance/elbow rather than computing a Pareto front or solving a constrained optimization problem. (polonsky2019threedimensionalanalysisand pages 5-6)\n\n**Assessment:** Area 3 is **partially occupied**. A broad claim to \u201coptimizing EBSD processing using grain-level outcomes\u201d would encounter Polonsky et al. A narrower architecture that jointly tunes indexing and cleanup parameters using several topology- and intervention-aware grain objectives still appears open in the searched literature.\n\n## 4. Hybrid Hough-first/dictionary-on-residue indexing\n\nThe closest automated escalation rule is Shen et al., but its first stage is a CNN rather than Hough indexing. Their proposed rule is explicit: high-confidence CNN results receive dictionary matching only near the preliminary orientation; low-confidence results receive full dictionary search. Thus, the publication anticipates confidence-triggered allocation of expensive pattern matching. It does not demonstrate a Hough-first pipeline, does not define residue using CI/IQ/MAD/NCC from Hough indexing, and does not report an executed end-to-end selective reindexing experiment. (shen2019convolutionalneuralnetworkbased pages 10-12)\n\nThe same paper compares CNN and Hough indexing at different exposure times. CNN indexing produced a larger fraction of points below a 10\u00b0 misorientation-error cutoff at high noise, whereas Hough was more accurate among the smaller subset it indexed correctly. This comparison motivates selective hybridization, but the reported hybrid suggestion remains CNN-to-dictionary rather than Hough-to-dictionary. (shen2019convolutionalneuralnetworkbased pages 8-10, shen2019convolutionalneuralnetworkbased pages 10-12)\n\nKaufmann et al. likewise replace conventional Hough phase assignment with CNN classification; they do not retain Hough as a first pass and selectively escalate failed pixels. (kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7)\n\n**Assessment:** Confidence-triggered hybrid search is neighboring prior art, so a very broad claim to \u201cuse confidence to decide whether to run dictionary matching\u201d is not open. However, no verified demonstration was found of **Hough first \u2192 automatically identify low-quality residue \u2192 dictionary/pattern-match only that residue**, particularly using an explicit CI/IQ/MAD/NCC or grain-context escalation criterion. Area 4 therefore remains mostly open in the requested Hough-specific form.\n\n## 5. Autonomous EBSD workflows and acquisition feedback\n\nSeveral publications automate substantial sub-workflows:\n\n* **BisQue\u2013DREAM.3D:** cloud execution, provenance, parallel pipeline runs, and range parameterization for 3D reconstruction. Parameters and ranges are supplied rather than autonomously discovered, and the demonstrated optimization begins after indexing. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6)\n* **Automated serial sectioning:** Burnett et al. describe a fully automated PFIB serial-section routine using automatically recognized tracking fiducials and repeated milling/imaging positions. Yet acceptable milling current was determined experimentally, and the authors state that systematic mapping of voltage/current space remained necessary. EBSD settings\u2014including beam energy/current, dwell, binning, band count, and Hough settings\u2014were fixed by the investigators rather than adapted from post-processing results. (burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7)\n* **Automated registration and distortion correction:** Sparks et al. use automated sample alignment, pattern-center fitting for each section, CI thresholding, and optimizer-based multimodal registration. The optimizer maximizes an image-registration similarity metric and transforms maps; it does not alter EBSD indexing settings or SEM acquisition conditions. (sparks20243dreconstructionof pages 9-12, sparks20243dreconstructionof pages 23-24)\n* **SE\u2013EBSD map correction:** Winiarski et al. automate aspects of co-registration and long-/short-range distortion correction using trainable filters and moving least squares. The paper explicitly concludes that more automated procedures and more robust accuracy measures still need development. It also discusses faster or smart acquisition as a prospect, not as a closed post-processing-to-acquisition controller. (winiarski2021correctionofartefacts pages 4-6, winiarski2021correctionofartefacts pages 15-17)\n* **CNN indexing/phase mapping:** Shen and Kaufmann automate inference from patterns, but training data, instrument calibration, phase standards, and/or model selection remain external; neither closes the loop to microscope settings. (shen2019convolutionalneuralnetworkbased pages 1-2, shen2019convolutionalneuralnetworkbased pages 8-10, kaufmann2020phasemappingin pages 1-2)\n\nNo retrieved work used post-processing outcomes to recommend or automatically change **beam current, pattern dwell/exposure, detector binning, or spatial step size** for a subsequent SEM/EBSD acquisition. Burnett et al. come closest to connecting acquisition conditions with indexing rate and band contrast, but the comparison is an investigator-designed experiment, not a controller: selected ion currents are evaluated under fixed EBSD conditions and an acceptable current is chosen experimentally. (burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7)\n\n**Assessment:** Automated components and scripted pipelines are well established, so \u201cautomated EBSD workflow\u201d alone is not a gap. A genuinely autonomous, human-out-of-the-loop EBSD characterization system that interprets indexed/map/grain quality and then recommends or controls SEM acquisition parameters remains open in the searched literature.\n\n## Consolidated gap assessment\n\n1. **Optimizer-driven tuning of EBSD pattern-processing/indexing settings \u2014 open.** Neighboring art covers per-pattern inverse optimization, ML indexing, and post-indexing segmentation sweeps, but not outer-loop optimization of background, binning, averaging, Hough, or dictionary hyperparameters.\n2. **Metric-driven iterative reindexing of stored raw patterns \u2014 open.** Raw-pattern reindexing is enabled and confidence-directed search is proposed, but the complete iterative objective loop was not found.\n3. **Grain-level optimization \u2014 partially claimed.** Polonsky et al. directly cover segmentation-tolerance selection using retained-good-data fraction and grain count. Broad multi-objective, constrained, topology-aware optimization across indexing plus cleanup remains open.\n4. **Automated hybrid escalation \u2014 partially anticipated, Hough-specific implementation open.** Shen et al. anticipate confidence-triggered local versus full dictionary search after CNN indexing. A demonstrated Hough-first, dictionary-on-residue workflow with formal escalation criteria was not found.\n5. **Autonomous workflow with feedback to acquisition \u2014 open at the system level.** Serial sectioning, registration, reconstruction, correction, and ML inference are individually automated, but no end-to-end post-processing-to-SEM acquisition feedback system was found.\n\nThe most defensible unexplored combination is therefore: **repeated reindexing of archived raw patterns under optimizer-selected processing/indexing settings; joint pattern-, pixel-, and grain-level objectives with explicit cleanup/topology constraints; confidence- and grain-context-based escalation from Hough to dictionary matching; and translation of the optimized result into acquisition recommendations or microscope control.** None of the verified publications combines those elements into a closed, autonomous EBSD loop.", + "raw_answer": "# Systematic prior-art survey: closed-loop optimization of EBSD processing and indexing\n\n## Scope and search interpretation\n\nThe survey distinguished four often-confounded categories: **(i)** optimization of pattern-processing or indexing parameters, **(ii)** algorithmic refinement of orientation or pattern-center estimates for a fixed pattern, **(iii)** automated post-indexing map cleanup/segmentation, and **(iv)** automation of acquisition or registration. Only the first category squarely satisfies area 1; automation in the other categories is relevant as neighboring prior art but does not establish a closed-loop EBSD indexing optimizer.\n\nSearches covered EBSD with Bayesian optimization, Gaussian processes, evolutionary/genetic algorithms, particle swarm optimization, grid/random search, reinforcement learning, Hough parameters, dictionary indexing, reindexing, CI/IQ/MAD/fit/NCC, grain-level objectives, autonomous microscopy, and acquisition feedback. Patent searches returned no relevant EBSD patent in the searched corpus. Absence findings below therefore mean **\u201cnot found in this systematic search,\u201d not a legal opinion of novelty or freedom to operate**.\n\n| Area | Closest published work(s) | What is automated | Objective/criterion | What is explicitly not automated | Gap assessment |\n|---|---|---|---|---|---|\n| 1. Optimizer-driven tuning of EBSD pattern-processing or indexing parameters | **Closest demonstrated EBSD parameter sweep:** Polonsky et al. 2019 parameterized DREAM.3D reconstruction variables and swept grain-segmentation misorientation tolerance over 0.1\u201310\u00b0 in 0.1\u00b0 increments, generating 100 reconstructions; **nearby but different:** Zhu et al. 2022 is only available here as metadata for \u201cglobal optimization algorithms\u201d applied to orientation/pattern-center/deformation extraction, so it should be treated as adjacent prior art, not verified from text in this record; **nearby non-tuning automation:** Kaufmann et al. 2020 CNN phase mapping automates phase recognition from patterns rather than tuning EBSD processing settings (pqac-00000014, pqac-00000013, pqac-00000004, pqac-00000005) | Automated batch execution / parameterization of reconstruction pipelines in BisQue+DREAM.3D; CNN-based direct phase classification from patterns in Kaufmann 2020 (pqac-00000014, pqac-00000004) | Polonsky used reconstruction-level metrics: fraction of \u201cgood\u201d CI-masked voxels retained and total grain count, selecting an elbow/balance point at 2.0\u00b0; Kaufmann targets phase-map fidelity qualitatively against known microstructure/EDS, not optimizer search (pqac-00000013, pqac-00000005) | No evidence here of Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, RL, random/grid search over **pattern** processing settings such as background correction, binning, neighbor averaging, Hough band count, rho-theta smoothing, or dictionary settings. Polonsky optimizes segmentation/reconstruction variables after indexing, not indexing settings themselves. Kaufmann does not tune EBSD parameters at all (pqac-00000013, pqac-00000005) | **Gap largely open.** Closest evidence is reconstruction-pipeline parameterization, not closed-loop optimizer-driven tuning of pattern-processing/indexing knobs. |\n| 2. Feedback loops that re-index stored raw patterns using CI/IQ/MAD/NCC to choose next parameter set | **Closest proposal:** Shen et al. 2019 proposes combining fast CNN predictions with a confidence measure (quaternion norm) so that high-confidence points get neighborhood dictionary search and low-confidence points get full dictionary search; **related raw-pattern availability, not loop:** Stinville et al. 2022 notes raw .up2 patterns can be opened in EMsoft DI / EMSphInx, enabling reindexing in principle (pqac-00000017, pqac-00000015, pqac-00000002, pqac-00000003) | Shen automates confidence estimation for each location and proposes routing subsequent dictionary matching effort based on confidence; raw pattern archival in Stinville enables offline reindexing workflows (pqac-00000017, pqac-00000003) | Quaternion norm as confidence proxy in Shen; no demonstrated CI/IQ/MAD/NCC-driven next-parameter selection loop in the gathered evidence (pqac-00000017) | Shen does **not** demonstrate an executed closed loop that iteratively changes indexing/processing parameters and re-indexes stored raw patterns to maximize CI/IQ/MAD/NCC/NCC-like objectives; it is a proposed hybrid search allocation strategy. Stinville provides data access, not autonomous optimization (pqac-00000017, pqac-00000003) | **Gap open.** Proposal-level hybrid confidence routing exists, but not a demonstrated feedback optimizer that selects the next EBSD parameter set from reindexed raw patterns. |\n| 3. Optimization against grain-level or microstructure-level outcomes | **Closest demonstrated:** Polonsky et al. 2019 explicitly optimizes a grain-segmentation parameter against microstructure-level outcomes (grain count and retained good-data fraction); Stan\u011bk et al. 2020 compares 2D/3D segmentation sensitivity to misorientation threshold and shows strong threshold dependence, but not autonomous optimization (pqac-00000013, pqac-00000012, pqac-00000010, pqac-00000011) | Automated parallel reconstructions and post-processing counts in Polonsky; analytical comparison of threshold effects on segmentation in Stan\u011bk (pqac-00000013, pqac-00000010) | Grain count stability / avoidance of grain merging plus retained CI-masked voxel fraction in Polonsky; sensitivity of number of grains and segmentation discrepancies vs threshold in Stan\u011bk (pqac-00000013, pqac-00000011) | No demonstrated multi-objective or constrained optimization over broader grain-level cleanup outcomes such as speckle/wild-grain fraction, grain-boundary integrity, fraction of pixels altered by cleanup, or Pareto balancing across multiple EBSD post-processing stages. Polonsky is a one-parameter sweep with manually interpreted elbow criterion (pqac-00000013, pqac-00000012) | **Partially occupied, but narrow.** Grain-level optimization exists for segmentation tolerance; broader autonomous multi-objective EBSD post-processing remains a gap. |\n| 4. Hybrid Hough-first / dictionary-on-residue and automated escalation criteria | **Closest published work:** Shen et al. 2019 explicitly proposes a hybrid strategy: CNN preliminary prediction plus confidence, then pattern matching only in a neighborhood for high-confidence points and full dictionary search for low-confidence points; this is the clearest automated escalation criterion in the gathered evidence. Sparks et al. 2024 and Winiarski et al. 2021 automate registration/distortion correction workflows but not Hough\u2192dictionary escalation (pqac-00000017, pqac-00000000, pqac-00000008) | Automated confidence scoring and proposed routing of dictionary-search scope in Shen; automated image registration/distortion correction in Sparks/Winiarski (pqac-00000017, pqac-00000000, pqac-00000008) | Criterion in Shen is confidence from quaternion norm: high confidence \u2192 local neighborhood dictionary search; low confidence \u2192 full dictionary search (pqac-00000017) | No evidence in the gathered text of a **demonstrated** Hough-first then dictionary-on-low-quality-residue loop with automated escalation based on CI/IQ/MAD/NCC thresholds. Shen\u2019s routing is proposed around CNN+dictionary, not Hough+dictionary, and not shown as an executed autonomous residue-reindexing pipeline (pqac-00000017) | **Mostly gap.** There is a close proposal for automated escalation, but not a demonstrated Hough-residue-to-dictionary autonomous workflow. |\n| 5. Autonomous/self-driving EBSD post-processing or characterization workflows, including links back to SEM acquisition recommendations | **Closest workflow automation:** Sparks et al. 2024 automates sample/image registration steps, pattern-center fitting for every section, and alignment routines; Winiarski et al. 2021 automates SE\u2013EBSD distortion correction and argues more automated procedures are needed; Burnett et al. 2016 demonstrates a fully automated serial-sectioning routine with tracking fiducials, but acquisition settings were experimentally chosen; Kaufmann et al. 2020 automates phase mapping from patterns (pqac-00000001, pqac-00000000, pqac-00000008, pqac-00000009, pqac-00000006, pqac-00000007, pqac-00000004) | Automated registration/warping, per-layer EBSD-to-BSE alignment, CI-threshold masks, pattern-center fitting per section, and automated serial sectioning with fiducial tracking; CNN inference for phase maps (pqac-00000000, pqac-00000001, pqac-00000006, pqac-00000004) | Workflow goals are geometric consistency, low inter-section misorientation, and corrected distortions; Burnett assesses equivalent quality via EBSD pattern quality/band contrast and indexing rate after experimentally selected settings (pqac-00000001, pqac-00000008, pqac-00000007) | No evidence here of a human-out-of-the-loop EBSD post-processing system that autonomously tunes post-processing/indexing parameters end-to-end, nor of a loop that feeds post-processing outcomes back to SEM acquisition parameters such as beam current, dwell time, or step size. Burnett explicitly determines acceptable milling current experimentally; Winiarski says more automated procedures still need to be developed (pqac-00000007, pqac-00000008) | **Gap open.** Sub-workflow automation exists, but not a self-driving closed-loop EBSD characterization system that recommends or controls acquisition parameters from post-processing outcomes. |\n\n\n*Table: This table maps the closest published prior art found for the five requested EBSD automation areas, distinguishing demonstrated methods from proposals and from absent closed loops. It highlights that most evidence clusters around workflow automation and grain-segmentation parameter sweeps, not autonomous closed-loop tuning of indexing/processing parameters.*\n\n## 1. Optimizer-driven tuning of pattern-processing or indexing parameters\n\n### Closest demonstrated work: Polonsky et al. (2019)\n\nPolonsky and colleagues integrated DREAM.3D into the cloud-based BisQue platform so that pipeline variables could be parameterized and executions parallelized. Their principal EBSD example varied the **post-indexing grain-segmentation misorientation tolerance** from 0.1\u00b0 to 10\u00b0 in 0.1\u00b0 increments, producing 100 three-dimensional reconstructions while holding minimum-neighbor and minimum-defect-size settings fixed. They evaluated the fraction of CI-masked \u201cgood\u201d voxels retained and total reconstructed grain count, then selected 2.0\u00b0 from convergence/elbow behavior in the derivatives of those metrics. This is genuine automated parameter sweeping with microstructure-level evaluation. However, it operates on already indexed orientation data; the study states that raw EBSD patterns were not collected for that dataset. It therefore does **not** tune background correction, detector binning, neighbor averaging, Hough band count, Hough rho/theta smoothing, or dictionary-search settings, and it cannot re-index patterns under candidate settings. (pqac-00000014, pqac-00000013, pqac-00000012)\n\n### Adjacent work that should not be overread\n\nShen et al. developed CNN-based orientation indexing and optimized neural-network architecture/training for speed and prediction accuracy. The trained model indexes patterns directly and uses quaternion-norm deviation as a reliability measure. This replaces or supplements conventional indexing; it does not run Bayesian, evolutionary, grid, or reinforcement-learning search over conventional EBSD processing parameters. It also remains sensitive to experimental geometry and requires detector/setup parameters to be determined beforehand or accommodated through retraining. (pqac-00000015, pqac-00000016, pqac-00000017)\n\nKaufmann et al. automated phase classification from individual EBSD patterns with a CNN. The work reduces human decisions in phase mapping and outperforms conventional Hough-plus-user-selected phase/chemistry workflows in difficult examples. The \u201coptimization\u201d described for the conventional comparator was repeated **user adjustment** of EDS reference chemistry against a microstructure already known in advance, not an algorithmic search over EBSD settings. The CNN likewise classifies patterns rather than optimizing background, Hough, binning, or dictionary parameters. (pqac-00000004, pqac-00000005)\n\nGlobal/evolutionary optimization has also been published for extracting or refining orientation, pattern center, and deformation state from pattern matching. That is important neighboring art: it establishes optimizer use **inside a per-pattern inverse problem**. It is not equivalent to an outer loop choosing data-processing/indexing hyperparameters from map-quality or grain-quality outcomes. The retrieved full-text evidence did not support treating such work as prior art for the broader closed-loop parameter-selection architecture.\n\n**Assessment:** No verified publication was found that applies Bayesian optimization, a GP surrogate, genetic/evolutionary search, random/grid search, RL, or a comparable optimizer to the listed **pattern-processing and indexing knobs** and automatically selects the best configuration. Area 1 remains substantially open, aside from adjacent per-pattern refinement and post-indexing segmentation sweeps.\n\n## 2. Re-indexing stored raw patterns in a metric-driven feedback loop\n\nStinville et al. provide a useful enabling precedent: their archived 3D dataset includes approximately 1.71 TB of raw `.up2` Kikuchi patterns, and the paper explicitly states that these can be opened and reprocessed with EMsoft dictionary indexing, EMSphInx spherical-pattern reindexing, or commercial EBSD software. The indexed files also expose IQ and CI. This demonstrates the technical availability of stored-pattern reindexing, but the paper is a dataset publication, not an optimizer: it does not generate candidate parameter sets, score them, or select the next experiment. (pqac-00000002, pqac-00000003)\n\nShen et al. provide the closest feedback-like decision rule. They propose obtaining a preliminary CNN orientation and confidence at every point, performing only a local dictionary search around high-confidence predictions, and performing a full dictionary search for low-confidence predictions. The confidence proxy is the norm of the predicted quaternion. This is automated computational routing, but it is not a demonstrated iterative loop over processing/indexing parameter sets, and it does not use CI, IQ, MAD/fit, or NCC to propose the next candidate configuration. (pqac-00000017)\n\nPolonsky et al. use CI to define an initial mask of good voxels and assess how many remain after cleanup and segmentation. CI is therefore part of a reconstruction-level quality assessment, but it is not fed back to re-index raw patterns; indeed, raw patterns were unavailable in that example. (pqac-00000013)\n\n**Assessment:** No demonstrated publication was found with the complete loop **candidate parameters \u2192 re-index stored raw patterns \u2192 compute CI/IQ/MAD/fit/NCC objective \u2192 choose the next parameter set \u2192 repeat**. Area 2 remains an unclaimed gap in the searched literature.\n\n## 3. Grain- or microstructure-level objectives\n\nThis area has the strongest direct prior art, but it is narrow.\n\nPolonsky et al. explicitly recognized that segmentation tolerance changes grain consolidation and retained data in opposite directions. Their two high-level metrics were total grain count and the fraction of initially CI-qualified voxels retained through reconstruction. They chose the lowest tolerance at which the changes in both quantities became approximately constant, analogous to an elbow criterion. This is a microstructure-level, effectively two-metric parameter-selection procedure rather than optimization of a per-pattern score. (pqac-00000013, pqac-00000012)\n\nStan\u011bk et al. independently show why such objectives matter. They analyzed the dependence of 2D and 3D grain segmentation on misorientation threshold, including mean grain counts and the formation of erroneously merged giant grains. For one noisy experimental dataset, changing noise removal and excluding grains below eight voxels changed the detected population from more than one million grains to 74,579. This is a systematic sensitivity study and supplies possible grain-level observables, but it does not implement an autonomous optimizer or select cleanup parameters through a formal multi-objective rule. (pqac-00000010, pqac-00000011)\n\nNeither publication demonstrates optimization across a broad EBSD-processing chain using objectives such as wild-grain/speckle fraction, boundary continuity or integrity, grain-count stability over threshold perturbations, fraction of pixels modified by cleanup, or a constraint on topology preservation. Polonsky varies essentially one segmentation parameter, keeps other settings fixed, and identifies a balance/elbow rather than computing a Pareto front or solving a constrained optimization problem. (pqac-00000013)\n\n**Assessment:** Area 3 is **partially occupied**. A broad claim to \u201coptimizing EBSD processing using grain-level outcomes\u201d would encounter Polonsky et al. A narrower architecture that jointly tunes indexing and cleanup parameters using several topology- and intervention-aware grain objectives still appears open in the searched literature.\n\n## 4. Hybrid Hough-first/dictionary-on-residue indexing\n\nThe closest automated escalation rule is Shen et al., but its first stage is a CNN rather than Hough indexing. Their proposed rule is explicit: high-confidence CNN results receive dictionary matching only near the preliminary orientation; low-confidence results receive full dictionary search. Thus, the publication anticipates confidence-triggered allocation of expensive pattern matching. It does not demonstrate a Hough-first pipeline, does not define residue using CI/IQ/MAD/NCC from Hough indexing, and does not report an executed end-to-end selective reindexing experiment. (pqac-00000017)\n\nThe same paper compares CNN and Hough indexing at different exposure times. CNN indexing produced a larger fraction of points below a 10\u00b0 misorientation-error cutoff at high noise, whereas Hough was more accurate among the smaller subset it indexed correctly. This comparison motivates selective hybridization, but the reported hybrid suggestion remains CNN-to-dictionary rather than Hough-to-dictionary. (pqac-00000016, pqac-00000017)\n\nKaufmann et al. likewise replace conventional Hough phase assignment with CNN classification; they do not retain Hough as a first pass and selectively escalate failed pixels. (pqac-00000004, pqac-00000005)\n\n**Assessment:** Confidence-triggered hybrid search is neighboring prior art, so a very broad claim to \u201cuse confidence to decide whether to run dictionary matching\u201d is not open. However, no verified demonstration was found of **Hough first \u2192 automatically identify low-quality residue \u2192 dictionary/pattern-match only that residue**, particularly using an explicit CI/IQ/MAD/NCC or grain-context escalation criterion. Area 4 therefore remains mostly open in the requested Hough-specific form.\n\n## 5. Autonomous EBSD workflows and acquisition feedback\n\nSeveral publications automate substantial sub-workflows:\n\n* **BisQue\u2013DREAM.3D:** cloud execution, provenance, parallel pipeline runs, and range parameterization for 3D reconstruction. Parameters and ranges are supplied rather than autonomously discovered, and the demonstrated optimization begins after indexing. (pqac-00000014, pqac-00000013)\n* **Automated serial sectioning:** Burnett et al. describe a fully automated PFIB serial-section routine using automatically recognized tracking fiducials and repeated milling/imaging positions. Yet acceptable milling current was determined experimentally, and the authors state that systematic mapping of voltage/current space remained necessary. EBSD settings\u2014including beam energy/current, dwell, binning, band count, and Hough settings\u2014were fixed by the investigators rather than adapted from post-processing results. (pqac-00000006, pqac-00000007)\n* **Automated registration and distortion correction:** Sparks et al. use automated sample alignment, pattern-center fitting for each section, CI thresholding, and optimizer-based multimodal registration. The optimizer maximizes an image-registration similarity metric and transforms maps; it does not alter EBSD indexing settings or SEM acquisition conditions. (pqac-00000001, pqac-00000000)\n* **SE\u2013EBSD map correction:** Winiarski et al. automate aspects of co-registration and long-/short-range distortion correction using trainable filters and moving least squares. The paper explicitly concludes that more automated procedures and more robust accuracy measures still need development. It also discusses faster or smart acquisition as a prospect, not as a closed post-processing-to-acquisition controller. (pqac-00000009, pqac-00000008)\n* **CNN indexing/phase mapping:** Shen and Kaufmann automate inference from patterns, but training data, instrument calibration, phase standards, and/or model selection remain external; neither closes the loop to microscope settings. (pqac-00000015, pqac-00000016, pqac-00000004)\n\nNo retrieved work used post-processing outcomes to recommend or automatically change **beam current, pattern dwell/exposure, detector binning, or spatial step size** for a subsequent SEM/EBSD acquisition. Burnett et al. come closest to connecting acquisition conditions with indexing rate and band contrast, but the comparison is an investigator-designed experiment, not a controller: selected ion currents are evaluated under fixed EBSD conditions and an acceptable current is chosen experimentally. (pqac-00000006, pqac-00000007)\n\n**Assessment:** Automated components and scripted pipelines are well established, so \u201cautomated EBSD workflow\u201d alone is not a gap. A genuinely autonomous, human-out-of-the-loop EBSD characterization system that interprets indexed/map/grain quality and then recommends or controls SEM acquisition parameters remains open in the searched literature.\n\n## Consolidated gap assessment\n\n1. **Optimizer-driven tuning of EBSD pattern-processing/indexing settings \u2014 open.** Neighboring art covers per-pattern inverse optimization, ML indexing, and post-indexing segmentation sweeps, but not outer-loop optimization of background, binning, averaging, Hough, or dictionary hyperparameters.\n2. **Metric-driven iterative reindexing of stored raw patterns \u2014 open.** Raw-pattern reindexing is enabled and confidence-directed search is proposed, but the complete iterative objective loop was not found.\n3. **Grain-level optimization \u2014 partially claimed.** Polonsky et al. directly cover segmentation-tolerance selection using retained-good-data fraction and grain count. Broad multi-objective, constrained, topology-aware optimization across indexing plus cleanup remains open.\n4. **Automated hybrid escalation \u2014 partially anticipated, Hough-specific implementation open.** Shen et al. anticipate confidence-triggered local versus full dictionary search after CNN indexing. A demonstrated Hough-first, dictionary-on-residue workflow with formal escalation criteria was not found.\n5. **Autonomous workflow with feedback to acquisition \u2014 open at the system level.** Serial sectioning, registration, reconstruction, correction, and ML inference are individually automated, but no end-to-end post-processing-to-SEM acquisition feedback system was found.\n\nThe most defensible unexplored combination is therefore: **repeated reindexing of archived raw patterns under optimizer-selected processing/indexing settings; joint pattern-, pixel-, and grain-level objectives with explicit cleanup/topology constraints; confidence- and grain-context-based escalation from Hough to dictionary matching; and translation of the optimized result into acquisition recommendations or microscope control.** None of the verified publications combines those elements into a closed, autonomous EBSD loop.", + "answer_reasoning": null, + "has_successful_answer": true, + "context": "", + "contexts": [ + { + "id": "pqac-3ae8e236", + "context": "The Shen et al. 2019 paper demonstrates CNN-based orientation indexing from EBSD patterns with computational efficiency (1 ms per pattern) and noise robustness. The method uses simulation-trained models that can be re-trained with limited experimental data. A quaternion norm metric is introduced to assess reconstruction quality and confidence without ground truth. The paper addresses the inverse problem of orientation reconstruction from diffraction patterns and shows the method avoids traditional image processing steps (Gaussian filtering, Hough transforms, peak detection). However, the excerpt does not explicitly discuss full dictionary search comparisons, low-confidence pattern handling strategies, iterative tuning of Hough/background/binning parameters, feedback loops to SEM acquisition, grain segmentation outcomes, or global/evolutionary optimization methods for pattern-center refinement.", + "question": "From the retrieved papers, extract exact evidence on: Shen et al. 2019 low-confidence patterns and full dictionary search; Polonsky et al. 2019 BisQue/DREAM.3D parameter sweeps and grain segmentation outcomes; global/evolutionary optimization in EBSD pattern matching or pattern-center/orientation refinement; and whether any method iteratively tunes Hough/background/binning/neighbor settings or closes feedback to SEM acquisition. State what is and is not automated.", + "text": { + "embedding": null, + "text": "", + "name": "shen2019convolutionalneuralnetworkbased pages 2-3", + "media": [], + "doc": { + "embedding": null, + "docname": "shen2019convolutionalneuralnetworkbased", + "dockey": "c7144b38ae859ed1", + "citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "shen2019convolutionalneuralnetworkbased", + "bibtex": "@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n", + "authors": [ + "Yu-Feng Shen", + "Reeju Pokharel", + "Thomas J. Nizolek", + "Anil Kumar", + "Turab Lookman" + ], + "publication_date": "2019-05-01T00:00:00Z", + "year": 2019, + "volume": "170", + "issue": null, + "issn": "1359-6454", + "pages": "118-131", + "journal": "Acta Materialia", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "title": "Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns", + "citation_count": 67, + "bibtex_type": "article", + "source_quality": 3, + "is_retracted": null, + "doi": "10.1016/j.actamat.2019.03.026", + "doi_url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "doc_id": "c7144b38ae859ed1", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1359645419301697", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal." + }, + "chunk_id": "bac7ce80-085b-4651-804c-3bec485ca97e" + }, + "score": 4, + "snippets": [ + "a simulation trained CNN to provide an initial estimate for re-training a small part of the network or combining it with an additional network with a small amount of experimental data results in high prediction accuracy for measured data", + "reconstructing a crystallographic orientation from an electron backscatter diffraction (EBSD) pattern is achieved by solving an inverse problem, i.e. finding the inverse function which maps a given diffraction pattern to an orientation: g(X) = q", + "The most commonly used approach for orientation indexing from measured EBSD patterns utilizes image processing techniques including Gaussian filtering, Hough transforms, butterfly convolution, and peak detection for extracting Kikuchi bands", + "a quaternion norm metric provides a good confidence measure for assessing the quality of the reconstruction, without requiring ground truth information", + "The CNN based reconstruction method takes as little as 1 ms for orientation indexing, matching the highest practical scan rates, and thus enabling real-time orientation indexing" + ], + "used_images": false + }, + { + "id": "pqac-d62fb3b4", + "context": "The Shen et al. 2019 paper discusses CNN-based orientation indexing for EBSD patterns. It addresses a key limitation: neural networks perform poorly on boundary points in the fundamental zone due to lack of crystallographic symmetry knowledge. The authors solve this by training multiple CNNs with transformed orientation spaces (effective fundamental zones). When using mean absolute error (MAE) loss, the network struggles because diffraction patterns from angles near boundaries (e.g., 0.001 and 1.999\u03c0) are visually similar but forced to predict very different values. The solution involves applying cubic symmetry operators to transform quaternions and create three additional effective fundamental zones where boundary points are repositioned to non-boundary locations, improving prediction accuracy. However, this excerpt does not contain information about low-confidence pattern handling, full dictionary search approaches, Polonsky et al. 2019 methods, evolutionary/global optimization techniques, or automated tuning of Hough parameters, background correction, binning settings, or feedback loops to SEM acquisition.", + "question": "From the retrieved papers, extract exact evidence on: Shen et al. 2019 low-confidence patterns and full dictionary search; Polonsky et al. 2019 BisQue/DREAM.3D parameter sweeps and grain segmentation outcomes; global/evolutionary optimization in EBSD pattern matching or pattern-center/orientation refinement; and whether any method iteratively tunes Hough/background/binning/neighbor settings or closes feedback to SEM acquisition. State what is and is not automated.", + "text": { + "embedding": null, + "text": "", + "name": "shen2019convolutionalneuralnetworkbased pages 13-14", + "media": [], + "doc": { + "embedding": null, + "docname": "shen2019convolutionalneuralnetworkbased", + "dockey": "c7144b38ae859ed1", + "citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "shen2019convolutionalneuralnetworkbased", + "bibtex": "@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n", + "authors": [ + "Yu-Feng Shen", + "Reeju Pokharel", + "Thomas J. Nizolek", + "Anil Kumar", + "Turab Lookman" + ], + "publication_date": "2019-05-01T00:00:00Z", + "year": 2019, + "volume": "170", + "issue": null, + "issn": "1359-6454", + "pages": "118-131", + "journal": "Acta Materialia", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "title": "Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns", + "citation_count": 67, + "bibtex_type": "article", + "source_quality": 3, + "is_retracted": null, + "doi": "10.1016/j.actamat.2019.03.026", + "doi_url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "doc_id": "c7144b38ae859ed1", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1359645419301697", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal." + }, + "chunk_id": "c2b6d66e-edfd-4165-bed8-459e093b8e43" + }, + "score": 3, + "snippets": [ + "the neural network performed poorly on boundary points, due to lack of knowledge about crystallographic symmetries", + "the diffraction patterns from q \u00bc 0:001 and q \u00bc 1:999p are very similar, but during the training process the neural network is forced to predict very different angle values for these diffraction patterns", + "training three additional CNNs with transformed orientations, while using the MAE loss function" + ], + "used_images": false + }, + { + "id": "pqac-00000014", + "context": "## Summary\n\nThis excerpt describes integration of DREAM.3D microstructure analysis software into BisQue, a cloud-based platform, to enable parallelized processing of 3D EBSD datasets from additively manufactured materials. The authors note that \"reconstruction of 3D datasets is a complex process with many degrees of freedom\" where \"algorithms employed, and tolerances selected by the user...ultimately determine the quantitative data extracted.\" They highlight that \"due to the long computation times required to process 3D data...it is impractical to explore the full parameter space by manually running reconstructions,\" and that \"only several variables are adjusted by a user until a qualitatively acceptable reconstruction is attained. This process inherently imparts bias.\" The work demonstrates parameterizing \"pipeline variables over a range of values\" to study grain segmentation misorientation tolerance, but focuses on cloud infrastructure rather than automated optimization of EBSD indexing parameters or closed-loop feedback mechanisms using quality metrics.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "e3d46ca7-0da2-467e-addc-e121db6ce344" + }, + "score": -1 + }, + { + "id": "pqac-00000010", + "context": "# Summary\n\nThis excerpt addresses grain segmentation in 2D and 3D EBSD measurements of polycrystalline materials. The authors present a comparison study using misorientation thresholds (typical value: 5\u00b0 in EDAX OIM software) to define grains as \"sets of all points (voxels or pixels) with similar orientation, surrounded by a boundary of a given misorientation threshold.\" A critical finding is that \"we may get different results of the segmentation in 2D and 3D, even if we use a fixed misorientation threshold.\" The paper uses \"Laguerre tessellations with orientation marks\" for simulation and employs \"DREAM 3D software\" for data evaluation. The discrepancy arises because \"in a 2D cutting plane, two sets of voxels may be detected as two grains, while in the 3D case, all these points belong to one larger set\" (particularly with nonconvex grains). The work demonstrates sensitivity differences between low-angle and high-angle grain boundary segmentation but does not address automated parameter optimization or feedback-loop indexing strategies.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "stanek2020comparisonofsegmentation pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "stanek2020comparisonofsegmentation", + "dockey": "c163ded6c3506dd4", + "citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stanek2020comparisonofsegmentation", + "bibtex": "@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n", + "authors": [ + "J. Stanek", + "J. Kope\u010dek", + "P. Kr\u00e1l", + "I. Karafi\u00e1tov\u00e1", + "F. Seitl", + "V. Bene\u0161" + ], + "publication_date": "2020-01-01T00:00:00Z", + "year": 2020, + "volume": "58", + "issue": "05", + "issn": "0023-432X", + "pages": "301-319", + "journal": "Metallic Materials", + "publisher": "Central Library of the Slovak Academy of Sciences", + "url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=f5jvks8o3ufok0s0abmso3h0r0", + "title": "Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.4149/km_2020_5_301", + "doi_url": "https://doi.org/10.4149/km_2020_5_301", + "doc_id": "c163ded6c3506dd4", + "file_location": null, + "license": null, + "pdf_url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=buc29d44pbljhmfhjpava1l817", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "crossref", + "semantic_scholar", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations." + }, + "chunk_id": "a9d9080d-189f-4444-80b2-98aea1dd69c4" + }, + "score": -1 + }, + { + "id": "pqac-00000004", + "context": "# Summary\n\nThis 2020 paper introduces convolutional neural networks (CNNs) for autonomous EBSD phase mapping, addressing limitations of conventional Hough transform-based indexing. The authors note that \"the Hough transform strips most of the information about the crystal structure and chemistry from the image\" and results in \"mis-classification of similar crystal structures.\" The work demonstrates CNN training to \"autonomously recognize the subtle differences in the diffraction patterns and output phase maps\" without extensive user input.\n\nThe paper acknowledges dictionary-based approaches (Park et al., 2013; Chen et al., 2015; Ram & De Graef, 2018) as solutions for phase differentiation, yet states these still rely on traditional Hough processing initially. The CNN methodology represents advancement toward \"high-throughput manner\" EBSD analysis with \"fewer human decisions,\" suggesting movement toward autonomous workflows, though the excerpt does not detail feedback loops, grain-level optimization metrics, or systematic parameter tuning strategies.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "kaufmann2020phasemappingin pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "kaufmann2020phasemappingin", + "dockey": "7fe5de81dfc311a6", + "citation": "Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "kaufmann2020phasemappingin", + "bibtex": "@article{kaufmann2020phasemappingin,\n author = \"Kaufmann, Kevin and Zhu, Chaoyi and Rosengarten, Alexander S. and Maryanovsky, Daniel and Wang, Haoren and Vecchio, Kenneth S.\",\n title = \"Phase Mapping in EBSD Using Convolutional Neural Networks\",\n year = \"2020\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"26\",\n pages = \"458-468\",\n month = \"May\",\n doi = \"10.1017/s1431927620001488\",\n url = \"https://doi.org/10.1017/s1431927620001488\",\n publisher = \"Oxford University Press (OUP)\",\n issue = \"3\",\n issn = \"1431-9276\"\n}\n", + "authors": [ + "Kevin Kaufmann", + "Chaoyi Zhu", + "Alexander S. Rosengarten", + "Daniel Maryanovsky", + "Haoren Wang", + "Kenneth S. Vecchio" + ], + "publication_date": "2020-05-11T00:00:00Z", + "year": 2020, + "volume": "26", + "issue": "3", + "issn": "1431-9276", + "pages": "458-468", + "journal": "Microscopy and Microanalysis", + "publisher": "Oxford University Press (OUP)", + "url": "https://www.cambridge.org/core/services/aop-cambridge-core/content/view/623282EB1DF6DB0559CEC8C74D30A74F/S1431927620001488a.pdf/div-class-title-phase-mapping-in-ebsd-using-convolutional-neural-networks-div.pdf", + "title": "Phase Mapping in EBSD Using Convolutional Neural Networks", + "citation_count": 34, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1017/s1431927620001488", + "doi_url": "https://doi.org/10.1017/s1431927620001488", + "doc_id": "7fe5de81dfc311a6", + "file_location": null, + "license": null, + "pdf_url": "https://academic.oup.com/mam/article-pdf/26/3/458/47996054/mam0458.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal." + }, + "chunk_id": "a458b081-cc56-413d-962a-a4e73624b9dd" + }, + "score": -1 + }, + { + "id": "pqac-00000001", + "context": "# Summary\n\nThis excerpt discusses comparison and alignment methodologies between EBSD and NF-HEDM crystallographic data, rather than automated optimization of EBSD parameters. Key findings include:\n\n- Layer-to-layer misorientation thresholds of 3\u00b0 used to define grain correspondence\n- NF-HEDM data showed uniformly small misorientation angles (all below 0.05\u00b0), while EBSD sections exceeded 0.3\u00b0 mean misorientation between sections\n- Automated sample alignment routines and pattern center fitting enabled high consistency\n- Per-layer empirical orientation alignment rotations reduced EBSD-HEDM misorientation: 95% of corresponding points achieved below 0.46\u00b0 after alignment (mean 0.24\u00b0)\n- Spherical harmonic indexing versus Hough indexing compared on raw EBSD patterns\n- 46.2% noncorresponding points between EBSD and HEDM data identified, attributed to smaller grains/twins absent in HEDM reconstructions\n\nThis excerpt addresses **data registration and post-processing comparison** but does not address automated parameter optimization, feedback loops using quality metrics, or grain-level outcome optimization.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "sparks20243dreconstructionof pages 9-12", + "media": [], + "doc": { + "embedding": null, + "docname": "sparks20243dreconstructionof", + "dockey": "be99dadb0f50f040", + "citation": "Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "sparks20243dreconstructionof", + "bibtex": "@article{sparks20243dreconstructionof,\n author = \"Sparks, Gregory and Mason, Simon A. and Chapman, Michael G. and Park, Jun-Sang and Sharma, Hemant and Kenesei, Peter and Niezgoda, Stephen R. and Mills, Michael J. and Uchic, Michael D. and Shade, Paul A. and Obstalecki, Mark\",\n title = \"3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jul\",\n doi = \"10.1007/s40192-024-00370-6\",\n url = \"https://doi.org/10.1007/s40192-024-00370-6\",\n volume = \"13\",\n pages = \"773-803\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"3\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Gregory Sparks", + "Simon A. Mason", + "Michael G. Chapman", + "Jun-Sang Park", + "Hemant Sharma", + "Peter Kenesei", + "Stephen R. Niezgoda", + "Michael J. Mills", + "Michael D. Uchic", + "Paul A. Shade", + "Mark Obstalecki" + ], + "publication_date": "2024-07-24T00:00:00Z", + "year": 2024, + "volume": "13", + "issue": "3", + "issn": "2193-9764", + "pages": "773-803", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-024-00370-6", + "title": "3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry", + "citation_count": 11, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-024-00370-6", + "doi_url": "https://doi.org/10.1007/s40192-024-00370-6", + "doc_id": "be99dadb0f50f040", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-024-00370-6.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal." + }, + "chunk_id": "fba05c57-83fd-4363-b638-4ac63c73d1e6" + }, + "score": -1 + }, + { + "id": "pqac-00000003", + "context": "# Summary\n\nThis excerpt describes EBSD data processing and validation from a TriBeam tomography study but does NOT address the question about automated optimization of EBSD parameters. The passage discusses:\n\n- **Data formats**: Raw Kikuchi patterns stored in \".up2\" binary files, processed using \"EMsoft dictionary indexing,\" \"EMSphInx spherical pattern reindexing,\" or commercial EDAX/Oxford software\n- **Quality metrics**: Image quality (IQ) and confidence index (CI) are mentioned as associated with collected patterns\n- **3D reconstruction**: Indexed data stored in HDF5 \".h5\" formatted files, visualized with ParaView\n- **Validation**: Grain boundary locations and crystallographic orientations verified by comparing conventional EBSD to TriBeam serial sectioning\n\nThe excerpt provides no evidence of automated parameter tuning, feedback loops, Bayesian optimization, genetic algorithms, grain-level optimization criteria, or autonomous workflows\u2014the core topics of the systematic survey question.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "stinville2022multimodaldatasetof pages 12-13", + "media": [], + "doc": { + "embedding": null, + "docname": "stinville2022multimodaldatasetof", + "dockey": "e97010568afd0695", + "citation": "J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stinville2022multimodaldatasetof", + "bibtex": "@article{stinville2022multimodaldatasetof,\n author = \"Stinville, J. C. and Hestroffer, J. M. and Charpagne, M. A. and Polonsky, A. T. and Echlin, M. P. and Torbet, C. J. and Valle, V. and Nygren, K. E. and Miller, M. P. and Klaas, O. and Loghin, A. and Beyerlein, I. J. and Pollock, T. M.\",\n title = \"Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields\",\n year = \"2022\",\n journal = \"Scientific Data\",\n volume = \"9\",\n month = \"Aug\",\n doi = \"10.1038/s41597-022-01525-w\",\n url = \"https://doi.org/10.1038/s41597-022-01525-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2052-4463\"\n}\n", + "authors": [ + "J. C. Stinville", + "J. M. Hestroffer", + "M. A. Charpagne", + "A. T. Polonsky", + "M. P. Echlin", + "C. J. Torbet", + "V. Valle", + "K. E. Nygren", + "M. P. Miller", + "O. Klaas", + "A. Loghin", + "I. J. Beyerlein", + "T. M. Pollock" + ], + "publication_date": "2022-08-01T00:00:00Z", + "year": 2022, + "volume": "9", + "issue": "1", + "issn": "2052-4463", + "pages": null, + "journal": "Scientific Data", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41597-022-01525-w", + "title": "Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields", + "citation_count": 36, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41597-022-01525-w", + "doi_url": "https://doi.org/10.1038/s41597-022-01525-w", + "doc_id": "e97010568afd0695", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41597-022-01525-w.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal." + }, + "chunk_id": "a6ff544c-e5e6-4418-9f8e-07f54adad212" + }, + "score": -1 + }, + { + "id": "pqac-00000005", + "context": "# Summary\n\nThe excerpt demonstrates CNN-based EBSD analysis outperforming traditional Hough transform indexing in multiphase materials. In a thermal barrier coating example with \"five FCC phases and one rhombohedral phase\" (all space group 225 except Al\u2082O\u2083), Hough-based EBSD \"unable to reliably differentiate the FCC phases\" produced incorrect phase fractions. Adding \"reference chemistry for each phase\" improved results but still showed errors\u2014\"Hastelloy X substrate has large grains and twins being identified as TaC.\" \n\nThe CNN-based model produced \"higher-fidelity mapping\" with phase fractions \"in good agreement with EDS maps\" and better spatial coherence. However, the excerpt does **not describe automated parameter optimization, feedback loops using quality metrics, grain-level objectives, hybrid indexing escalation criteria, or autonomous post-processing workflows**\u2014instead focusing on pattern classification accuracy using pre-trained neural networks versus user-dependent Hough-transform library selection.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "kaufmann2020phasemappingin pages 6-7", + "media": [], + "doc": { + "embedding": null, + "docname": "kaufmann2020phasemappingin", + "dockey": "7fe5de81dfc311a6", + "citation": "Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "kaufmann2020phasemappingin", + "bibtex": "@article{kaufmann2020phasemappingin,\n author = \"Kaufmann, Kevin and Zhu, Chaoyi and Rosengarten, Alexander S. and Maryanovsky, Daniel and Wang, Haoren and Vecchio, Kenneth S.\",\n title = \"Phase Mapping in EBSD Using Convolutional Neural Networks\",\n year = \"2020\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"26\",\n pages = \"458-468\",\n month = \"May\",\n doi = \"10.1017/s1431927620001488\",\n url = \"https://doi.org/10.1017/s1431927620001488\",\n publisher = \"Oxford University Press (OUP)\",\n issue = \"3\",\n issn = \"1431-9276\"\n}\n", + "authors": [ + "Kevin Kaufmann", + "Chaoyi Zhu", + "Alexander S. Rosengarten", + "Daniel Maryanovsky", + "Haoren Wang", + "Kenneth S. Vecchio" + ], + "publication_date": "2020-05-11T00:00:00Z", + "year": 2020, + "volume": "26", + "issue": "3", + "issn": "1431-9276", + "pages": "458-468", + "journal": "Microscopy and Microanalysis", + "publisher": "Oxford University Press (OUP)", + "url": "https://www.cambridge.org/core/services/aop-cambridge-core/content/view/623282EB1DF6DB0559CEC8C74D30A74F/S1431927620001488a.pdf/div-class-title-phase-mapping-in-ebsd-using-convolutional-neural-networks-div.pdf", + "title": "Phase Mapping in EBSD Using Convolutional Neural Networks", + "citation_count": 34, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1017/s1431927620001488", + "doi_url": "https://doi.org/10.1017/s1431927620001488", + "doc_id": "7fe5de81dfc311a6", + "file_location": null, + "license": null, + "pdf_url": "https://academic.oup.com/mam/article-pdf/26/3/458/47996054/mam0458.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal." + }, + "chunk_id": "20a0778b-e502-4592-9f60-4cef4d99a32b" + }, + "score": -1 + }, + { + "id": "pqac-00000002", + "context": "# Summary\n\nThis excerpt describes post-processing of EBSD data from an In718 sample, including voxelization, geometric reconstruction, and meshing workflows. Key details include:\n\n- Raw EBSD data provided in multiple formats: IPF maps (tiff, RGB), crystallographic orientation data (.osc, .ang formats with Euler angles \u03c61, \u03a6, \u03c62), and raw Kikuchi patterns (.up2 binary format)\n- EBSD-derived quality metrics reported: \"image quality (IQ) and confidence index (CI) associated with the collected Kikuchi patterns\"\n- Post-processing includes artifact removal (features <50 voxels), erosion/dilation (3\u00d73\u00d73 block), and smoothing of grain boundaries\n- References EMsoft \"dictionary indexing\" and \"EMSphInx spherical pattern reindexing\" tools for pattern processing\n- No discussion of automated parameter optimization, feedback loops, grain-level objective functions, hybrid indexing criteria, or autonomous workflow frameworks\n\n**Gap assessment**: The excerpt does not address any of the five specified areas regarding automated/optimized EBSD parameter tuning or autonomous workflows.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "stinville2022multimodaldatasetof pages 9-12", + "media": [], + "doc": { + "embedding": null, + "docname": "stinville2022multimodaldatasetof", + "dockey": "e97010568afd0695", + "citation": "J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stinville2022multimodaldatasetof", + "bibtex": "@article{stinville2022multimodaldatasetof,\n author = \"Stinville, J. C. and Hestroffer, J. M. and Charpagne, M. A. and Polonsky, A. T. and Echlin, M. P. and Torbet, C. J. and Valle, V. and Nygren, K. E. and Miller, M. P. and Klaas, O. and Loghin, A. and Beyerlein, I. J. and Pollock, T. M.\",\n title = \"Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields\",\n year = \"2022\",\n journal = \"Scientific Data\",\n volume = \"9\",\n month = \"Aug\",\n doi = \"10.1038/s41597-022-01525-w\",\n url = \"https://doi.org/10.1038/s41597-022-01525-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2052-4463\"\n}\n", + "authors": [ + "J. C. Stinville", + "J. M. Hestroffer", + "M. A. Charpagne", + "A. T. Polonsky", + "M. P. Echlin", + "C. J. Torbet", + "V. Valle", + "K. E. Nygren", + "M. P. Miller", + "O. Klaas", + "A. Loghin", + "I. J. Beyerlein", + "T. M. Pollock" + ], + "publication_date": "2022-08-01T00:00:00Z", + "year": 2022, + "volume": "9", + "issue": "1", + "issn": "2052-4463", + "pages": null, + "journal": "Scientific Data", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41597-022-01525-w", + "title": "Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields", + "citation_count": 36, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41597-022-01525-w", + "doi_url": "https://doi.org/10.1038/s41597-022-01525-w", + "doc_id": "e97010568afd0695", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41597-022-01525-w.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal." + }, + "chunk_id": "4b2ea9a0-4239-4bf2-97a7-a4352930daa3" + }, + "score": -1 + }, + { + "id": "pqac-00000016", + "context": "# Summary\n\nThis excerpt discusses CNN-based EBSD pattern indexing from Shen et al. (2019), comparing it to Hough transform and dictionary methods. Key findings:\n\n**What it does:** Uses U-Net and CNN models to predict crystal orientations from EBSD patterns, achieving \"50% of all points having <10\u00b0 misorientation\" at 1ms exposure versus \"~15%\" for Hough transform. The U-Net reconstructed 1\u00d71 mm\u00b2 scans in \"17 min (4 ms per pattern)\" on a MacBook Pro; retrained CNN in \"5 min (1 ms per pattern).\"\n\n**What it does NOT automate:** The CNN requires \"re-training the weights\" when experimental geometry parameters change. The authors note that \"a single neural network will not provide accurate predictions for all experimental geometry parameters.\" They suggest combining CNN with dictionary indexing but do not present automated criteria for method selection.\n\n**Gap identified:** The excerpt proposes but does not implement \"combining the two approaches: first apply CNN-based regression to obtain pre-[liminary results]\"\u2014indicating no published automated hybrid indexing strategy exists in this work.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "shen2019convolutionalneuralnetworkbased pages 8-10", + "media": [], + "doc": { + "embedding": null, + "docname": "shen2019convolutionalneuralnetworkbased", + "dockey": "c7144b38ae859ed1", + "citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "shen2019convolutionalneuralnetworkbased", + "bibtex": "@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n", + "authors": [ + "Yu-Feng Shen", + "Reeju Pokharel", + "Thomas J. Nizolek", + "Anil Kumar", + "Turab Lookman" + ], + "publication_date": "2019-05-01T00:00:00Z", + "year": 2019, + "volume": "170", + "issue": null, + "issn": "1359-6454", + "pages": "118-131", + "journal": "Acta Materialia", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "title": "Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns", + "citation_count": 67, + "bibtex_type": "article", + "source_quality": 3, + "is_retracted": null, + "doi": "10.1016/j.actamat.2019.03.026", + "doi_url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "doc_id": "c7144b38ae859ed1", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1359645419301697", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal." + }, + "chunk_id": "a0708b6a-0b0d-4604-9033-ddf735241059" + }, + "score": -1 + }, + { + "id": "pqac-00000013", + "context": "# Summary\n\nThis excerpt describes a parameterization study of EBSD data post-processing, specifically optimizing the **misorientation tolerance parameter** for 3D grain segmentation. Key details:\n\n- **Grain-level optimization metrics**: The work tracks two opposing trends\u2014\"number of grains\" and \"fraction of masked voxels\"\u2014as misorientation tolerance varies from 0.1\u00b0 to 10\u00b0 in 0.1\u00b0 increments, generating \"one-hundred 3D reconstructions.\"\n\n- **Optimization method**: Uses the \"Elbow method\" via central difference derivatives to find the point where \"both the change in fraction of masked data included and the change in the number of grains became constant,\" identifying 2.0\u00b0 as optimal.\n\n- **Microstructure-level focus**: Explicitly addresses avoiding \"unnecessary merging of grains\" and ensuring \"grains with more reasonable grain morphologies,\" prioritizing grain-boundary integrity over per-pixel pattern scores.\n\n- **Tool used**: BisQue framework enables automated parameterization and multi-reconstruction analysis.\n\n- **Automation scope**: Parameter tuning is semi-automated but does NOT employ Bayesian optimization, genetic algorithms, or closed-loop feedback on raw EBSD patterns.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 5-6", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "4425c958-1a47-4afc-bdd5-8e5ab6ab1538" + }, + "score": -1 + }, + { + "id": "pqac-00000011", + "context": "# Summary\n\nThis excerpt describes grain segmentation analysis using 3D-EBSD data on an AlMgSc alloy. The study examines how misorientation thresholds affect grain detection when comparing 2D versus 3D processing approaches. Key findings include:\n\n- Three misorientation thresholds tested: 0.1\u00b0, 10%, and 15% of boundaries\n- When the specimen thickness is \u226530 slices, 2D vs 3D processing differences become \"significant\"\n- The difference becomes meaningful when \"at least 10\u201315% of boundaries have misorientation lower than the threshold\"\n- Grain morphology plays minimal role; sample thickness is the main factor\n- Processing parameters \"firmly influence the number of detected grains\" \u2014 using DREAM.3D with a 2\u00b0 misorientation threshold detected 74,579 grains, but without noise filtering detected >1 million grains\n- Noise substantially affects results; removing grains <8 voxels and applying filters is critical\n\nThe excerpt demonstrates **grain-level outcome variability** (grain count) dependent on processing parameter selection but does **not** describe automated optimization loops or feedback mechanisms for parameter tuning.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "stanek2020comparisonofsegmentation pages 7-9", + "media": [], + "doc": { + "embedding": null, + "docname": "stanek2020comparisonofsegmentation", + "dockey": "c163ded6c3506dd4", + "citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stanek2020comparisonofsegmentation", + "bibtex": "@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n", + "authors": [ + "J. Stanek", + "J. Kope\u010dek", + "P. Kr\u00e1l", + "I. Karafi\u00e1tov\u00e1", + "F. Seitl", + "V. Bene\u0161" + ], + "publication_date": "2020-01-01T00:00:00Z", + "year": 2020, + "volume": "58", + "issue": "05", + "issn": "0023-432X", + "pages": "301-319", + "journal": "Metallic Materials", + "publisher": "Central Library of the Slovak Academy of Sciences", + "url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=f5jvks8o3ufok0s0abmso3h0r0", + "title": "Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.4149/km_2020_5_301", + "doi_url": "https://doi.org/10.4149/km_2020_5_301", + "doc_id": "c163ded6c3506dd4", + "file_location": null, + "license": null, + "pdf_url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=buc29d44pbljhmfhjpava1l817", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "crossref", + "semantic_scholar", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations." + }, + "chunk_id": "86978fee-1fa5-4882-aa4a-b780c83742b6" + }, + "score": -1 + }, + { + "id": "pqac-00000015", + "context": "# Summary\n\nThis excerpt presents a CNN-based approach for EBSD orientation indexing by Shen et al. (2019). Key points:\n\n**What is automated:** The paper develops \"convolutional neural network approaches based on domain transform and transfer learning to reconstruct crystal orientations from electron backscatter diffraction patterns.\" The CNN models \"index orientations as fast as the highest EBSD scanning rates\" and are \"robust to experimentally measured image noise.\"\n\n**Quality assessment:** The authors \"demonstrate that the quaternion norm metric is a strong indicator for assessing the reliability of the reconstructions.\"\n\n**Limitations:** The CNN approach addresses orientation indexing speed but does not explicitly automate parameter tuning, feedback loops, grain-level optimization, hybrid indexing strategies, or closed-loop SEM acquisition control. Prior CNN work (Liu et al., Jha et al.) was \"only trained and tested on simulated EBSD patterns,\" though this work extends to experimental data using \"transfer learning\" with \"a small amount of experimental data.\"\n\nThis represents machine learning for pattern indexing but not systematic parameter optimization or autonomous workflow control.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "shen2019convolutionalneuralnetworkbased pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "shen2019convolutionalneuralnetworkbased", + "dockey": "c7144b38ae859ed1", + "citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "shen2019convolutionalneuralnetworkbased", + "bibtex": "@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n", + "authors": [ + "Yu-Feng Shen", + "Reeju Pokharel", + "Thomas J. Nizolek", + "Anil Kumar", + "Turab Lookman" + ], + "publication_date": "2019-05-01T00:00:00Z", + "year": 2019, + "volume": "170", + "issue": null, + "issn": "1359-6454", + "pages": "118-131", + "journal": "Acta Materialia", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "title": "Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns", + "citation_count": 67, + "bibtex_type": "article", + "source_quality": 3, + "is_retracted": null, + "doi": "10.1016/j.actamat.2019.03.026", + "doi_url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "doc_id": "c7144b38ae859ed1", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1359645419301697", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal." + }, + "chunk_id": "c9ca9e15-4082-449b-b4d2-5375c2a351d0" + }, + "score": -1 + }, + { + "id": "pqac-00000007", + "context": "This excerpt discusses EBSD (electron backscatter diffraction) analysis of WC-Co samples prepared with different ion-beam milling techniques, comparing results across BIB (Ar+), PFIB (Xe+), and FIB (Ga+) methods at varying currents and voltages.\n\nKey findings: BIB at 6 kV produced \"very high quality\" EBSD patterns with 90% WC indexing and band contrast of 180/255. PFIB at 30 kV, 59 nA achieved 86% indexing with band contrast 108/255. FIB at 20 nA showed poor results (8% indexing, 40/255 band contrast), while FIB at 1 nA improved substantially (76% indexing, 94/255 contrast). The excerpt notes that \"60\u00d7 higher current (1 nA vs. 59 nA)\" could be used in FIB versus PFIB to achieve \"roughly equivalent quality\" based on band contrast and indexing metrics. The authors acknowledge \"considerable systematic work is required to map this parameter space\" but do not describe automated optimization approaches or feedback loops for parameter selection.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "burnett2016largevolumeserial pages 6-7", + "media": [], + "doc": { + "embedding": null, + "docname": "burnett2016largevolumeserial", + "dockey": "a2e8eb201167feb2", + "citation": "Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "burnett2016largevolumeserial", + "bibtex": "@article{burnett2016largevolumeserial,\n author = \"Burnett, Timothy L. and Burnett, Timothy L. and Kelley, R. and Winiarski, B. and Contreras, L. and Daly, Michael and Gholinia, A. and Burke, M. G. and Withers, P. J.\",\n title = \"Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.\",\n year = \"2016\",\n journal = \"Ultramicroscopy\",\n volume = \"161\",\n pages = \"119-129\",\n month = \"Feb\",\n doi = \"10.1016/j.ultramic.2015.11.001\",\n url = \"https://doi.org/10.1016/j.ultramic.2015.11.001\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n", + "authors": [ + "Timothy L. Burnett", + "Timothy L. Burnett", + "R. Kelley", + "B. Winiarski", + "L. Contreras", + "Michael Daly", + "A. Gholinia", + "M. G. Burke", + "P. J. Withers" + ], + "publication_date": "2016-02-01T00:00:00Z", + "year": 2016, + "volume": "161", + "issue": null, + "issn": "0304-3991", + "pages": "119-129", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2015.11.001", + "title": "Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.", + "citation_count": 440, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2015.11.001", + "doi_url": "https://doi.org/10.1016/j.ultramic.2015.11.001", + "doc_id": "a2e8eb201167feb2", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/pii/S0304399115300644/pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal." + }, + "chunk_id": "e499cab0-252a-4ffb-a321-e89ba6ed941b" + }, + "score": -1 + }, + { + "id": "pqac-00000008", + "context": "# Summary\n\nThis excerpt addresses **3D-EBSD data correction**, not automated parameter optimization or indexing. It describes a methodology for correcting geometric distortions in EBSD maps using \"trainable Combination of Shifted Filter Responses filters and two-dimensional Moving Least Squares method\" applied to WC-Co hard metals.\n\nKey findings:\n- Quantified \"long-range distortions...of the order of up to 10 \u03bcm across a 200 \u00d7 200 \u03bcm area\"\n- Long-range corrections reduced planar facet topography from 3.3 \u03bcm to 2.9 \u03bcm\n- Short-range corrections further reduced roughness to 0.4 \u03bcm\n- The authors acknowledge that \"more automated procedures need to be developed so that large area EBSD maps can be routinely corrected\"\n\n**Relevance to survey**: This work does NOT address automated tuning of EBSD pattern-processing parameters, feedback loops with quality metrics, grain-level optimization objectives, hybrid indexing strategies, or autonomous workflows. It focuses narrowly on post-acquisition geometric correction, leaving all five survey areas unaddressed.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "winiarski2021correctionofartefacts pages 15-17", + "media": [], + "doc": { + "embedding": null, + "docname": "winiarski2021correctionofartefacts", + "dockey": "5e305820c91848ff", + "citation": "B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "winiarski2021correctionofartefacts", + "bibtex": "@misc{winiarski2021correctionofartefacts,\n author = \"Winiarski, B. and Gholinia, A. and Mingard, K. and Gee, M. and Thompson, G. and Withers, P.J.\",\n title = \"Correction of artefacts associated with large area EBSD\",\n year = \"2021\",\n journal = \"Ultramicroscopy\",\n volume = \"226\",\n pages = \"113315\",\n month = \"Jul\",\n doi = \"10.1016/j.ultramic.2021.113315\",\n url = \"https://doi.org/10.1016/j.ultramic.2021.113315\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n", + "authors": [ + "B. Winiarski", + "A. Gholinia", + "K. Mingard", + "M. Gee", + "G. Thompson", + "P.J. Withers" + ], + "publication_date": "2021-07-01T00:00:00Z", + "year": 2021, + "volume": "226", + "issue": null, + "issn": "0304-3991", + "pages": "113315", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2021.113315", + "title": "Correction of artefacts associated with large area EBSD", + "citation_count": 30, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2021.113315", + "doi_url": "https://doi.org/10.1016/j.ultramic.2021.113315", + "doc_id": "5e305820c91848ff", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/pii/S0304399121001017", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal." + }, + "chunk_id": "afefd03f-e33c-4e7a-8fcb-9c9face205c7" + }, + "score": -1 + }, + { + "id": "pqac-00000000", + "context": "# Summary\n\nThis excerpt describes image registration and processing workflows for multi-modal microscopy data (optical, BSE, EBSD, SWLI), but **does not address any of the five specified areas** regarding automated EBSD parameter optimization.\n\nThe text details:\n- Use of SITK framework with B-Spline transforms and mutual information similarity metrics for spatial registration\n- ORIX Python suite for IPF colormap generation and confidence index-based masking\n- Manual selection of hyperparameters (mesh size 1\u00d71 to 3\u00d73, B-Spline order 2, L-BFGS-B optimizer)\n- Post-processing of EBSD orientation data after registration\n\nNotably absent: no automated parameter tuning, no feedback loops using quality metrics (CI, IQ, NCC) as objective functions, no multi-objective optimization against grain-level outcomes, no hybrid indexing strategies, and no autonomous workflows. The work performs downstream data fusion and registration but leaves EBSD indexing parameters and processing entirely unaddressed\u2014confirming these remain literature gaps.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "sparks20243dreconstructionof pages 23-24", + "media": [], + "doc": { + "embedding": null, + "docname": "sparks20243dreconstructionof", + "dockey": "be99dadb0f50f040", + "citation": "Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "sparks20243dreconstructionof", + "bibtex": "@article{sparks20243dreconstructionof,\n author = \"Sparks, Gregory and Mason, Simon A. and Chapman, Michael G. and Park, Jun-Sang and Sharma, Hemant and Kenesei, Peter and Niezgoda, Stephen R. and Mills, Michael J. and Uchic, Michael D. and Shade, Paul A. and Obstalecki, Mark\",\n title = \"3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jul\",\n doi = \"10.1007/s40192-024-00370-6\",\n url = \"https://doi.org/10.1007/s40192-024-00370-6\",\n volume = \"13\",\n pages = \"773-803\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"3\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Gregory Sparks", + "Simon A. Mason", + "Michael G. Chapman", + "Jun-Sang Park", + "Hemant Sharma", + "Peter Kenesei", + "Stephen R. Niezgoda", + "Michael J. Mills", + "Michael D. Uchic", + "Paul A. Shade", + "Mark Obstalecki" + ], + "publication_date": "2024-07-24T00:00:00Z", + "year": 2024, + "volume": "13", + "issue": "3", + "issn": "2193-9764", + "pages": "773-803", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-024-00370-6", + "title": "3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry", + "citation_count": 11, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-024-00370-6", + "doi_url": "https://doi.org/10.1007/s40192-024-00370-6", + "doc_id": "be99dadb0f50f040", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-024-00370-6.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal." + }, + "chunk_id": "7f431cc5-d62d-4dea-86ef-1f6af04bf4c0" + }, + "score": -1 + }, + { + "id": "pqac-00000006", + "context": "# Summary\n\nThis excerpt describes serial sectioning methodology and EBSD acquisition parameters for microstructural analysis, but does **not address the systematic survey question asked**.\n\nThe text details:\n- **Sample preparation**: FIB milling procedures using specific currents (1.3 \u03bcA to 59 nA) and voltages (30 kV)\n- **EBSD acquisition parameters**: 20 kV beam energy, 4 nA current, step sizes of 50\u201375 nm, dwell times of 2.3\u20136 ms, \"4\u00d74 binning, 6\u201312 bands, RA and Hough=50\u201360\"\n- **Image processing**: Alignment using least-squares algorithms, FFT filtering, watershed/top-hat segmentation in Avizo 9.0.0\n- **Automated milling routine**: Rocking polish sequence with fiducial tracking\n\n**Critical gap**: The excerpt provides *fixed, predetermined* EBSD parameters and describes *semi-automated segmentation*, but contains **no evidence of automated parameter optimization, feedback loops, quality-metric-driven indexing, grain-level optimization, hybrid indexing strategies, or autonomous workflow frameworks**\u2014the five areas specified in the survey question.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "burnett2016largevolumeserial pages 4-6", + "media": [], + "doc": { + "embedding": null, + "docname": "burnett2016largevolumeserial", + "dockey": "a2e8eb201167feb2", + "citation": "Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "burnett2016largevolumeserial", + "bibtex": "@article{burnett2016largevolumeserial,\n author = \"Burnett, Timothy L. and Burnett, Timothy L. and Kelley, R. and Winiarski, B. and Contreras, L. and Daly, Michael and Gholinia, A. and Burke, M. G. and Withers, P. J.\",\n title = \"Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.\",\n year = \"2016\",\n journal = \"Ultramicroscopy\",\n volume = \"161\",\n pages = \"119-129\",\n month = \"Feb\",\n doi = \"10.1016/j.ultramic.2015.11.001\",\n url = \"https://doi.org/10.1016/j.ultramic.2015.11.001\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n", + "authors": [ + "Timothy L. Burnett", + "Timothy L. Burnett", + "R. Kelley", + "B. Winiarski", + "L. Contreras", + "Michael Daly", + "A. Gholinia", + "M. G. Burke", + "P. J. Withers" + ], + "publication_date": "2016-02-01T00:00:00Z", + "year": 2016, + "volume": "161", + "issue": null, + "issn": "0304-3991", + "pages": "119-129", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2015.11.001", + "title": "Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.", + "citation_count": 440, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2015.11.001", + "doi_url": "https://doi.org/10.1016/j.ultramic.2015.11.001", + "doc_id": "a2e8eb201167feb2", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/pii/S0304399115300644/pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal." + }, + "chunk_id": "984c5e66-1791-4cb5-8c20-7e837aaa4348" + }, + "score": -1 + }, + { + "id": "pqac-00000012", + "context": "# Summary\n\nThe excerpt describes parameter optimization work on 3D EBSD data processing using the BisQue framework and DREAM.3D software. Key findings:\n\n**Automated parameter tuning**: The authors systematically tested misorientation tolerance values from 0.1\u00b0 to 10\u00b0 (100 increments) on 3D reconstructions, using \"two high-level metrics...the fraction of 'good' data, and the total number of grains found.\" They identified an optimal misorientation tolerance of 2.0\u00b0 through derivative analysis balancing grain count stability and data retention (98.4% of masked data included, 899 grains identified).\n\n**Grain-level outcomes**: The work explicitly optimizes against grain-level metrics rather than per-pixel scores\u2014measuring fraction of retained data and total grain counts as reconstruction quality indicators.\n\n**Coupled parameterization**: BisQue enables \"automated generation of random samples of equal-sized subvolumes,\" allowing linked parameter sets.\n\nHowever, the excerpt does **not** demonstrate: automated feedback loops re-indexing raw patterns, Bayesian/genetic algorithm optimization, hybrid indexing strategies, or autonomous SEM acquisition parameter recommendations.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 6-8", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "7ead8b49-300a-4424-92e3-3c9cede9a7d5" + }, + "score": -1 + }, + { + "id": "pqac-00000009", + "context": "# Summary\n\nThe excerpt discusses challenges in EBSD data acquisition and processing, particularly regarding spatial distortions in large-area maps. Key points:\n\n**EBSD Acquisition Challenges:**\n- EBSD requires \"one to three orders of magnitude longer\" dwell times than SE imaging, resulting in \"acquisition times of hundreds of minutes per map\"\n- Maps acquired at 70\u00b0 stage tilt experience \"1/cosine(70\u00b0) or 2.9 times smaller beam spacing,\" causing \"almost a threefold degradation in spatial resolution\"\n- Long-range and short-range distortions are \"largely neglected at present\"\n\n**Emerging Solutions Mentioned:**\n- \"Dictionary based indexing\" and \"smart acquisition\" promise \"faster acquisition times without compromising EBSD indexing rates\"\n- Distortion correction methods using COSFIRE filters and Moving Least Squares (MLS) deformation are proposed\n\n**Experimental Details:**\nThe WC-Co hardmetal case study used 200 nm step size, 4\u00d74 binning (336\u00d7256 pixels), 25 ms indexing per point (40 Hz), achieving \">96%\" indexing rate\u2014notably slower than typical operation.\n\nThe excerpt emphasizes distortion as an understudied problem but does not address automated parameter optimization, feedback loops, or grain-level outcome optimization.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "winiarski2021correctionofartefacts pages 4-6", + "media": [], + "doc": { + "embedding": null, + "docname": "winiarski2021correctionofartefacts", + "dockey": "5e305820c91848ff", + "citation": "B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "winiarski2021correctionofartefacts", + "bibtex": "@misc{winiarski2021correctionofartefacts,\n author = \"Winiarski, B. and Gholinia, A. and Mingard, K. and Gee, M. and Thompson, G. and Withers, P.J.\",\n title = \"Correction of artefacts associated with large area EBSD\",\n year = \"2021\",\n journal = \"Ultramicroscopy\",\n volume = \"226\",\n pages = \"113315\",\n month = \"Jul\",\n doi = \"10.1016/j.ultramic.2021.113315\",\n url = \"https://doi.org/10.1016/j.ultramic.2021.113315\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n", + "authors": [ + "B. Winiarski", + "A. Gholinia", + "K. Mingard", + "M. Gee", + "G. Thompson", + "P.J. Withers" + ], + "publication_date": "2021-07-01T00:00:00Z", + "year": 2021, + "volume": "226", + "issue": null, + "issn": "0304-3991", + "pages": "113315", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2021.113315", + "title": "Correction of artefacts associated with large area EBSD", + "citation_count": 30, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2021.113315", + "doi_url": "https://doi.org/10.1016/j.ultramic.2021.113315", + "doc_id": "5e305820c91848ff", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/pii/S0304399121001017", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal." + }, + "chunk_id": "f79c497e-55f5-4e4b-bbd0-fe90be825530" + }, + "score": -1 + }, + { + "id": "pqac-00000017", + "context": "# Summary\n\nThe excerpt describes CNN-based approaches for EBSD pattern indexing and microstructure reconstruction. Key points:\n\n**CNN Method:** Demonstrated \"relatively fast\" real-time reconstruction capability during scanning, but \"not as accurate as the dictionary approach, which can reconstruct noisy data with high accuracy.\"\n\n**Hybrid Strategy:** Proposes combining CNN and dictionary methods: \"first apply CNN-based regression to obtain preliminary predictions and a measure of confidence for each location...For locations with high confidence reconstructions, run the pattern matching only in the neighborhood (e.g. 10+) of the preliminary results in the dictionary. For those with low confidence, run the full dictionary search.\"\n\n**Confidence Metric:** Uses \"quaternion norm as a measure of confidence\" to assess reconstruction reliability and determine which indexing approach to apply.\n\n**Domain Adaptation:** Employs U-Net and transfer learning to adapt CNNs trained on simulated patterns to experimental data, addressing simulation-experiment discrepancies through convolutional layer retraining.\n\nThe work addresses noise robustness and matching \"EBSD scanning rates\" but focuses on per-pattern indexing rather than grain-level optimization or automated parameter tuning.", + "question": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "text": { + "embedding": null, + "text": "", + "name": "shen2019convolutionalneuralnetworkbased pages 10-12", + "media": [], + "doc": { + "embedding": null, + "docname": "shen2019convolutionalneuralnetworkbased", + "dockey": "c7144b38ae859ed1", + "citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "shen2019convolutionalneuralnetworkbased", + "bibtex": "@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n", + "authors": [ + "Yu-Feng Shen", + "Reeju Pokharel", + "Thomas J. Nizolek", + "Anil Kumar", + "Turab Lookman" + ], + "publication_date": "2019-05-01T00:00:00Z", + "year": 2019, + "volume": "170", + "issue": null, + "issn": "1359-6454", + "pages": "118-131", + "journal": "Acta Materialia", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "title": "Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns", + "citation_count": 67, + "bibtex_type": "article", + "source_quality": 3, + "is_retracted": null, + "doi": "10.1016/j.actamat.2019.03.026", + "doi_url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "doc_id": "c7144b38ae859ed1", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1359645419301697", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal." + }, + "chunk_id": "246b0249-4200-4f19-a695-3c50ef9b7730" + }, + "score": -1 + } + ], + "references": "1. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n2. (polonsky2019threedimensionalanalysisand pages 5-6): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n3. (kaufmann2020phasemappingin pages 1-2): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal.\n\n4. (kaufmann2020phasemappingin pages 6-7): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal.\n\n5. (shen2019convolutionalneuralnetworkbased pages 10-12): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal.\n\n6. (shen2019convolutionalneuralnetworkbased pages 1-2): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal.\n\n7. (stinville2022multimodaldatasetof pages 9-12): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal.\n\n8. (stinville2022multimodaldatasetof pages 12-13): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal.\n\n9. (polonsky2019threedimensionalanalysisand pages 6-8): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n10. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\n11. (stanek2020comparisonofsegmentation pages 7-9): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\n12. (sparks20243dreconstructionof pages 23-24): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal.\n\n13. (winiarski2021correctionofartefacts pages 15-17): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal.\n\n14. (sparks20243dreconstructionof pages 9-12): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal.\n\n15. (winiarski2021correctionofartefacts pages 4-6): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal.\n\n16. (burnett2016largevolumeserial pages 4-6): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal.\n\n17. (burnett2016largevolumeserial pages 6-7): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal.\n\n18. (shen2019convolutionalneuralnetworkbased pages 8-10): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal.", + "formatted_answer": "Question: Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.\n\n# Systematic prior-art survey: closed-loop optimization of EBSD processing and indexing\n\n## Scope and search interpretation\n\nThe survey distinguished four often-confounded categories: **(i)** optimization of pattern-processing or indexing parameters, **(ii)** algorithmic refinement of orientation or pattern-center estimates for a fixed pattern, **(iii)** automated post-indexing map cleanup/segmentation, and **(iv)** automation of acquisition or registration. Only the first category squarely satisfies area 1; automation in the other categories is relevant as neighboring prior art but does not establish a closed-loop EBSD indexing optimizer.\n\nSearches covered EBSD with Bayesian optimization, Gaussian processes, evolutionary/genetic algorithms, particle swarm optimization, grid/random search, reinforcement learning, Hough parameters, dictionary indexing, reindexing, CI/IQ/MAD/fit/NCC, grain-level objectives, autonomous microscopy, and acquisition feedback. Patent searches returned no relevant EBSD patent in the searched corpus. Absence findings below therefore mean **\u201cnot found in this systematic search,\u201d not a legal opinion of novelty or freedom to operate**.\n\n| Area | Closest published work(s) | What is automated | Objective/criterion | What is explicitly not automated | Gap assessment |\n|---|---|---|---|---|---|\n| 1. Optimizer-driven tuning of EBSD pattern-processing or indexing parameters | **Closest demonstrated EBSD parameter sweep:** Polonsky et al. 2019 parameterized DREAM.3D reconstruction variables and swept grain-segmentation misorientation tolerance over 0.1\u201310\u00b0 in 0.1\u00b0 increments, generating 100 reconstructions; **nearby but different:** Zhu et al. 2022 is only available here as metadata for \u201cglobal optimization algorithms\u201d applied to orientation/pattern-center/deformation extraction, so it should be treated as adjacent prior art, not verified from text in this record; **nearby non-tuning automation:** Kaufmann et al. 2020 CNN phase mapping automates phase recognition from patterns rather than tuning EBSD processing settings (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7) | Automated batch execution / parameterization of reconstruction pipelines in BisQue+DREAM.3D; CNN-based direct phase classification from patterns in Kaufmann 2020 (polonsky2019threedimensionalanalysisand pages 1-2, kaufmann2020phasemappingin pages 1-2) | Polonsky used reconstruction-level metrics: fraction of \u201cgood\u201d CI-masked voxels retained and total grain count, selecting an elbow/balance point at 2.0\u00b0; Kaufmann targets phase-map fidelity qualitatively against known microstructure/EDS, not optimizer search (polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 6-7) | No evidence here of Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, RL, random/grid search over **pattern** processing settings such as background correction, binning, neighbor averaging, Hough band count, rho-theta smoothing, or dictionary settings. Polonsky optimizes segmentation/reconstruction variables after indexing, not indexing settings themselves. Kaufmann does not tune EBSD parameters at all (polonsky2019threedimensionalanalysisand pages 5-6, kaufmann2020phasemappingin pages 6-7) | **Gap largely open.** Closest evidence is reconstruction-pipeline parameterization, not closed-loop optimizer-driven tuning of pattern-processing/indexing knobs. |\n| 2. Feedback loops that re-index stored raw patterns using CI/IQ/MAD/NCC to choose next parameter set | **Closest proposal:** Shen et al. 2019 proposes combining fast CNN predictions with a confidence measure (quaternion norm) so that high-confidence points get neighborhood dictionary search and low-confidence points get full dictionary search; **related raw-pattern availability, not loop:** Stinville et al. 2022 notes raw .up2 patterns can be opened in EMsoft DI / EMSphInx, enabling reindexing in principle (shen2019convolutionalneuralnetworkbased pages 10-12, shen2019convolutionalneuralnetworkbased pages 1-2, stinville2022multimodaldatasetof pages 9-12, stinville2022multimodaldatasetof pages 12-13) | Shen automates confidence estimation for each location and proposes routing subsequent dictionary matching effort based on confidence; raw pattern archival in Stinville enables offline reindexing workflows (shen2019convolutionalneuralnetworkbased pages 10-12, stinville2022multimodaldatasetof pages 12-13) | Quaternion norm as confidence proxy in Shen; no demonstrated CI/IQ/MAD/NCC-driven next-parameter selection loop in the gathered evidence (shen2019convolutionalneuralnetworkbased pages 10-12) | Shen does **not** demonstrate an executed closed loop that iteratively changes indexing/processing parameters and re-indexes stored raw patterns to maximize CI/IQ/MAD/NCC/NCC-like objectives; it is a proposed hybrid search allocation strategy. Stinville provides data access, not autonomous optimization (shen2019convolutionalneuralnetworkbased pages 10-12, stinville2022multimodaldatasetof pages 12-13) | **Gap open.** Proposal-level hybrid confidence routing exists, but not a demonstrated feedback optimizer that selects the next EBSD parameter set from reindexed raw patterns. |\n| 3. Optimization against grain-level or microstructure-level outcomes | **Closest demonstrated:** Polonsky et al. 2019 explicitly optimizes a grain-segmentation parameter against microstructure-level outcomes (grain count and retained good-data fraction); Stan\u011bk et al. 2020 compares 2D/3D segmentation sensitivity to misorientation threshold and shows strong threshold dependence, but not autonomous optimization (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8, stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 7-9) | Automated parallel reconstructions and post-processing counts in Polonsky; analytical comparison of threshold effects on segmentation in Stan\u011bk (polonsky2019threedimensionalanalysisand pages 5-6, stanek2020comparisonofsegmentation pages 1-2) | Grain count stability / avoidance of grain merging plus retained CI-masked voxel fraction in Polonsky; sensitivity of number of grains and segmentation discrepancies vs threshold in Stan\u011bk (polonsky2019threedimensionalanalysisand pages 5-6, stanek2020comparisonofsegmentation pages 7-9) | No demonstrated multi-objective or constrained optimization over broader grain-level cleanup outcomes such as speckle/wild-grain fraction, grain-boundary integrity, fraction of pixels altered by cleanup, or Pareto balancing across multiple EBSD post-processing stages. Polonsky is a one-parameter sweep with manually interpreted elbow criterion (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8) | **Partially occupied, but narrow.** Grain-level optimization exists for segmentation tolerance; broader autonomous multi-objective EBSD post-processing remains a gap. |\n| 4. Hybrid Hough-first / dictionary-on-residue and automated escalation criteria | **Closest published work:** Shen et al. 2019 explicitly proposes a hybrid strategy: CNN preliminary prediction plus confidence, then pattern matching only in a neighborhood for high-confidence points and full dictionary search for low-confidence points; this is the clearest automated escalation criterion in the gathered evidence. Sparks et al. 2024 and Winiarski et al. 2021 automate registration/distortion correction workflows but not Hough\u2192dictionary escalation (shen2019convolutionalneuralnetworkbased pages 10-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17) | Automated confidence scoring and proposed routing of dictionary-search scope in Shen; automated image registration/distortion correction in Sparks/Winiarski (shen2019convolutionalneuralnetworkbased pages 10-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17) | Criterion in Shen is confidence from quaternion norm: high confidence \u2192 local neighborhood dictionary search; low confidence \u2192 full dictionary search (shen2019convolutionalneuralnetworkbased pages 10-12) | No evidence in the gathered text of a **demonstrated** Hough-first then dictionary-on-low-quality-residue loop with automated escalation based on CI/IQ/MAD/NCC thresholds. Shen\u2019s routing is proposed around CNN+dictionary, not Hough+dictionary, and not shown as an executed autonomous residue-reindexing pipeline (shen2019convolutionalneuralnetworkbased pages 10-12) | **Mostly gap.** There is a close proposal for automated escalation, but not a demonstrated Hough-residue-to-dictionary autonomous workflow. |\n| 5. Autonomous/self-driving EBSD post-processing or characterization workflows, including links back to SEM acquisition recommendations | **Closest workflow automation:** Sparks et al. 2024 automates sample/image registration steps, pattern-center fitting for every section, and alignment routines; Winiarski et al. 2021 automates SE\u2013EBSD distortion correction and argues more automated procedures are needed; Burnett et al. 2016 demonstrates a fully automated serial-sectioning routine with tracking fiducials, but acquisition settings were experimentally chosen; Kaufmann et al. 2020 automates phase mapping from patterns (sparks20243dreconstructionof pages 9-12, sparks20243dreconstructionof pages 23-24, winiarski2021correctionofartefacts pages 15-17, winiarski2021correctionofartefacts pages 4-6, burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7, kaufmann2020phasemappingin pages 1-2) | Automated registration/warping, per-layer EBSD-to-BSE alignment, CI-threshold masks, pattern-center fitting per section, and automated serial sectioning with fiducial tracking; CNN inference for phase maps (sparks20243dreconstructionof pages 23-24, sparks20243dreconstructionof pages 9-12, burnett2016largevolumeserial pages 4-6, kaufmann2020phasemappingin pages 1-2) | Workflow goals are geometric consistency, low inter-section misorientation, and corrected distortions; Burnett assesses equivalent quality via EBSD pattern quality/band contrast and indexing rate after experimentally selected settings (sparks20243dreconstructionof pages 9-12, winiarski2021correctionofartefacts pages 15-17, burnett2016largevolumeserial pages 6-7) | No evidence here of a human-out-of-the-loop EBSD post-processing system that autonomously tunes post-processing/indexing parameters end-to-end, nor of a loop that feeds post-processing outcomes back to SEM acquisition parameters such as beam current, dwell time, or step size. Burnett explicitly determines acceptable milling current experimentally; Winiarski says more automated procedures still need to be developed (burnett2016largevolumeserial pages 6-7, winiarski2021correctionofartefacts pages 15-17) | **Gap open.** Sub-workflow automation exists, but not a self-driving closed-loop EBSD characterization system that recommends or controls acquisition parameters from post-processing outcomes. |\n\n\n*Table: This table maps the closest published prior art found for the five requested EBSD automation areas, distinguishing demonstrated methods from proposals and from absent closed loops. It highlights that most evidence clusters around workflow automation and grain-segmentation parameter sweeps, not autonomous closed-loop tuning of indexing/processing parameters.*\n\n## 1. Optimizer-driven tuning of pattern-processing or indexing parameters\n\n### Closest demonstrated work: Polonsky et al. (2019)\n\nPolonsky and colleagues integrated DREAM.3D into the cloud-based BisQue platform so that pipeline variables could be parameterized and executions parallelized. Their principal EBSD example varied the **post-indexing grain-segmentation misorientation tolerance** from 0.1\u00b0 to 10\u00b0 in 0.1\u00b0 increments, producing 100 three-dimensional reconstructions while holding minimum-neighbor and minimum-defect-size settings fixed. They evaluated the fraction of CI-masked \u201cgood\u201d voxels retained and total reconstructed grain count, then selected 2.0\u00b0 from convergence/elbow behavior in the derivatives of those metrics. This is genuine automated parameter sweeping with microstructure-level evaluation. However, it operates on already indexed orientation data; the study states that raw EBSD patterns were not collected for that dataset. It therefore does **not** tune background correction, detector binning, neighbor averaging, Hough band count, Hough rho/theta smoothing, or dictionary-search settings, and it cannot re-index patterns under candidate settings. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8)\n\n### Adjacent work that should not be overread\n\nShen et al. developed CNN-based orientation indexing and optimized neural-network architecture/training for speed and prediction accuracy. The trained model indexes patterns directly and uses quaternion-norm deviation as a reliability measure. This replaces or supplements conventional indexing; it does not run Bayesian, evolutionary, grid, or reinforcement-learning search over conventional EBSD processing parameters. It also remains sensitive to experimental geometry and requires detector/setup parameters to be determined beforehand or accommodated through retraining. (shen2019convolutionalneuralnetworkbased pages 1-2, shen2019convolutionalneuralnetworkbased pages 8-10, shen2019convolutionalneuralnetworkbased pages 10-12)\n\nKaufmann et al. automated phase classification from individual EBSD patterns with a CNN. The work reduces human decisions in phase mapping and outperforms conventional Hough-plus-user-selected phase/chemistry workflows in difficult examples. The \u201coptimization\u201d described for the conventional comparator was repeated **user adjustment** of EDS reference chemistry against a microstructure already known in advance, not an algorithmic search over EBSD settings. The CNN likewise classifies patterns rather than optimizing background, Hough, binning, or dictionary parameters. (kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7)\n\nGlobal/evolutionary optimization has also been published for extracting or refining orientation, pattern center, and deformation state from pattern matching. That is important neighboring art: it establishes optimizer use **inside a per-pattern inverse problem**. It is not equivalent to an outer loop choosing data-processing/indexing hyperparameters from map-quality or grain-quality outcomes. The retrieved full-text evidence did not support treating such work as prior art for the broader closed-loop parameter-selection architecture.\n\n**Assessment:** No verified publication was found that applies Bayesian optimization, a GP surrogate, genetic/evolutionary search, random/grid search, RL, or a comparable optimizer to the listed **pattern-processing and indexing knobs** and automatically selects the best configuration. Area 1 remains substantially open, aside from adjacent per-pattern refinement and post-indexing segmentation sweeps.\n\n## 2. Re-indexing stored raw patterns in a metric-driven feedback loop\n\nStinville et al. provide a useful enabling precedent: their archived 3D dataset includes approximately 1.71 TB of raw `.up2` Kikuchi patterns, and the paper explicitly states that these can be opened and reprocessed with EMsoft dictionary indexing, EMSphInx spherical-pattern reindexing, or commercial EBSD software. The indexed files also expose IQ and CI. This demonstrates the technical availability of stored-pattern reindexing, but the paper is a dataset publication, not an optimizer: it does not generate candidate parameter sets, score them, or select the next experiment. (stinville2022multimodaldatasetof pages 9-12, stinville2022multimodaldatasetof pages 12-13)\n\nShen et al. provide the closest feedback-like decision rule. They propose obtaining a preliminary CNN orientation and confidence at every point, performing only a local dictionary search around high-confidence predictions, and performing a full dictionary search for low-confidence predictions. The confidence proxy is the norm of the predicted quaternion. This is automated computational routing, but it is not a demonstrated iterative loop over processing/indexing parameter sets, and it does not use CI, IQ, MAD/fit, or NCC to propose the next candidate configuration. (shen2019convolutionalneuralnetworkbased pages 10-12)\n\nPolonsky et al. use CI to define an initial mask of good voxels and assess how many remain after cleanup and segmentation. CI is therefore part of a reconstruction-level quality assessment, but it is not fed back to re-index raw patterns; indeed, raw patterns were unavailable in that example. (polonsky2019threedimensionalanalysisand pages 5-6)\n\n**Assessment:** No demonstrated publication was found with the complete loop **candidate parameters \u2192 re-index stored raw patterns \u2192 compute CI/IQ/MAD/fit/NCC objective \u2192 choose the next parameter set \u2192 repeat**. Area 2 remains an unclaimed gap in the searched literature.\n\n## 3. Grain- or microstructure-level objectives\n\nThis area has the strongest direct prior art, but it is narrow.\n\nPolonsky et al. explicitly recognized that segmentation tolerance changes grain consolidation and retained data in opposite directions. Their two high-level metrics were total grain count and the fraction of initially CI-qualified voxels retained through reconstruction. They chose the lowest tolerance at which the changes in both quantities became approximately constant, analogous to an elbow criterion. This is a microstructure-level, effectively two-metric parameter-selection procedure rather than optimization of a per-pattern score. (polonsky2019threedimensionalanalysisand pages 5-6, polonsky2019threedimensionalanalysisand pages 6-8)\n\nStan\u011bk et al. independently show why such objectives matter. They analyzed the dependence of 2D and 3D grain segmentation on misorientation threshold, including mean grain counts and the formation of erroneously merged giant grains. For one noisy experimental dataset, changing noise removal and excluding grains below eight voxels changed the detected population from more than one million grains to 74,579. This is a systematic sensitivity study and supplies possible grain-level observables, but it does not implement an autonomous optimizer or select cleanup parameters through a formal multi-objective rule. (stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 7-9)\n\nNeither publication demonstrates optimization across a broad EBSD-processing chain using objectives such as wild-grain/speckle fraction, boundary continuity or integrity, grain-count stability over threshold perturbations, fraction of pixels modified by cleanup, or a constraint on topology preservation. Polonsky varies essentially one segmentation parameter, keeps other settings fixed, and identifies a balance/elbow rather than computing a Pareto front or solving a constrained optimization problem. (polonsky2019threedimensionalanalysisand pages 5-6)\n\n**Assessment:** Area 3 is **partially occupied**. A broad claim to \u201coptimizing EBSD processing using grain-level outcomes\u201d would encounter Polonsky et al. A narrower architecture that jointly tunes indexing and cleanup parameters using several topology- and intervention-aware grain objectives still appears open in the searched literature.\n\n## 4. Hybrid Hough-first/dictionary-on-residue indexing\n\nThe closest automated escalation rule is Shen et al., but its first stage is a CNN rather than Hough indexing. Their proposed rule is explicit: high-confidence CNN results receive dictionary matching only near the preliminary orientation; low-confidence results receive full dictionary search. Thus, the publication anticipates confidence-triggered allocation of expensive pattern matching. It does not demonstrate a Hough-first pipeline, does not define residue using CI/IQ/MAD/NCC from Hough indexing, and does not report an executed end-to-end selective reindexing experiment. (shen2019convolutionalneuralnetworkbased pages 10-12)\n\nThe same paper compares CNN and Hough indexing at different exposure times. CNN indexing produced a larger fraction of points below a 10\u00b0 misorientation-error cutoff at high noise, whereas Hough was more accurate among the smaller subset it indexed correctly. This comparison motivates selective hybridization, but the reported hybrid suggestion remains CNN-to-dictionary rather than Hough-to-dictionary. (shen2019convolutionalneuralnetworkbased pages 8-10, shen2019convolutionalneuralnetworkbased pages 10-12)\n\nKaufmann et al. likewise replace conventional Hough phase assignment with CNN classification; they do not retain Hough as a first pass and selectively escalate failed pixels. (kaufmann2020phasemappingin pages 1-2, kaufmann2020phasemappingin pages 6-7)\n\n**Assessment:** Confidence-triggered hybrid search is neighboring prior art, so a very broad claim to \u201cuse confidence to decide whether to run dictionary matching\u201d is not open. However, no verified demonstration was found of **Hough first \u2192 automatically identify low-quality residue \u2192 dictionary/pattern-match only that residue**, particularly using an explicit CI/IQ/MAD/NCC or grain-context escalation criterion. Area 4 therefore remains mostly open in the requested Hough-specific form.\n\n## 5. Autonomous EBSD workflows and acquisition feedback\n\nSeveral publications automate substantial sub-workflows:\n\n* **BisQue\u2013DREAM.3D:** cloud execution, provenance, parallel pipeline runs, and range parameterization for 3D reconstruction. Parameters and ranges are supplied rather than autonomously discovered, and the demonstrated optimization begins after indexing. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 5-6)\n* **Automated serial sectioning:** Burnett et al. describe a fully automated PFIB serial-section routine using automatically recognized tracking fiducials and repeated milling/imaging positions. Yet acceptable milling current was determined experimentally, and the authors state that systematic mapping of voltage/current space remained necessary. EBSD settings\u2014including beam energy/current, dwell, binning, band count, and Hough settings\u2014were fixed by the investigators rather than adapted from post-processing results. (burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7)\n* **Automated registration and distortion correction:** Sparks et al. use automated sample alignment, pattern-center fitting for each section, CI thresholding, and optimizer-based multimodal registration. The optimizer maximizes an image-registration similarity metric and transforms maps; it does not alter EBSD indexing settings or SEM acquisition conditions. (sparks20243dreconstructionof pages 9-12, sparks20243dreconstructionof pages 23-24)\n* **SE\u2013EBSD map correction:** Winiarski et al. automate aspects of co-registration and long-/short-range distortion correction using trainable filters and moving least squares. The paper explicitly concludes that more automated procedures and more robust accuracy measures still need development. It also discusses faster or smart acquisition as a prospect, not as a closed post-processing-to-acquisition controller. (winiarski2021correctionofartefacts pages 4-6, winiarski2021correctionofartefacts pages 15-17)\n* **CNN indexing/phase mapping:** Shen and Kaufmann automate inference from patterns, but training data, instrument calibration, phase standards, and/or model selection remain external; neither closes the loop to microscope settings. (shen2019convolutionalneuralnetworkbased pages 1-2, shen2019convolutionalneuralnetworkbased pages 8-10, kaufmann2020phasemappingin pages 1-2)\n\nNo retrieved work used post-processing outcomes to recommend or automatically change **beam current, pattern dwell/exposure, detector binning, or spatial step size** for a subsequent SEM/EBSD acquisition. Burnett et al. come closest to connecting acquisition conditions with indexing rate and band contrast, but the comparison is an investigator-designed experiment, not a controller: selected ion currents are evaluated under fixed EBSD conditions and an acceptable current is chosen experimentally. (burnett2016largevolumeserial pages 4-6, burnett2016largevolumeserial pages 6-7)\n\n**Assessment:** Automated components and scripted pipelines are well established, so \u201cautomated EBSD workflow\u201d alone is not a gap. A genuinely autonomous, human-out-of-the-loop EBSD characterization system that interprets indexed/map/grain quality and then recommends or controls SEM acquisition parameters remains open in the searched literature.\n\n## Consolidated gap assessment\n\n1. **Optimizer-driven tuning of EBSD pattern-processing/indexing settings \u2014 open.** Neighboring art covers per-pattern inverse optimization, ML indexing, and post-indexing segmentation sweeps, but not outer-loop optimization of background, binning, averaging, Hough, or dictionary hyperparameters.\n2. **Metric-driven iterative reindexing of stored raw patterns \u2014 open.** Raw-pattern reindexing is enabled and confidence-directed search is proposed, but the complete iterative objective loop was not found.\n3. **Grain-level optimization \u2014 partially claimed.** Polonsky et al. directly cover segmentation-tolerance selection using retained-good-data fraction and grain count. Broad multi-objective, constrained, topology-aware optimization across indexing plus cleanup remains open.\n4. **Automated hybrid escalation \u2014 partially anticipated, Hough-specific implementation open.** Shen et al. anticipate confidence-triggered local versus full dictionary search after CNN indexing. A demonstrated Hough-first, dictionary-on-residue workflow with formal escalation criteria was not found.\n5. **Autonomous workflow with feedback to acquisition \u2014 open at the system level.** Serial sectioning, registration, reconstruction, correction, and ML inference are individually automated, but no end-to-end post-processing-to-SEM acquisition feedback system was found.\n\nThe most defensible unexplored combination is therefore: **repeated reindexing of archived raw patterns under optimizer-selected processing/indexing settings; joint pattern-, pixel-, and grain-level objectives with explicit cleanup/topology constraints; confidence- and grain-context-based escalation from Hough to dictionary matching; and translation of the optimized result into acquisition recommendations or microscope control.** None of the verified publications combines those elements into a closed, autonomous EBSD loop.\n\nReferences\n\n1. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n2. (polonsky2019threedimensionalanalysisand pages 5-6): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n3. (kaufmann2020phasemappingin pages 1-2): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal.\n\n4. (kaufmann2020phasemappingin pages 6-7): Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal.\n\n5. (shen2019convolutionalneuralnetworkbased pages 10-12): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal.\n\n6. (shen2019convolutionalneuralnetworkbased pages 1-2): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal.\n\n7. (stinville2022multimodaldatasetof pages 9-12): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal.\n\n8. (stinville2022multimodaldatasetof pages 12-13): J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal.\n\n9. (polonsky2019threedimensionalanalysisand pages 6-8): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n10. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\n11. (stanek2020comparisonofsegmentation pages 7-9): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\n12. (sparks20243dreconstructionof pages 23-24): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal.\n\n13. (winiarski2021correctionofartefacts pages 15-17): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal.\n\n14. (sparks20243dreconstructionof pages 9-12): Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal.\n\n15. (winiarski2021correctionofartefacts pages 4-6): B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal.\n\n16. (burnett2016largevolumeserial pages 4-6): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal.\n\n17. (burnett2016largevolumeserial pages 6-7): Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal.\n\n18. (shen2019convolutionalneuralnetworkbased pages 8-10): Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal.", + "graded_answer": null, + "cost": 0.0, + "token_counts": {}, + "config_md5": "44fbe82f4172525021b58c01857cebcc", + "tool_history": [ + [ + "create_plan" + ], + [ + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "patent_search", + "patent_search" + ], + [ + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search" + ], + [ + "gather_evidence" + ], + [ + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search" + ], + [ + "gather_evidence" + ], + [ + "read_text", + "read_text", + "read_text" + ], + [ + "update_plan" + ], + [ + "update_plan" + ], + [ + "create_artifact" + ], + [ + "update_plan", + "update_plan", + "update_plan", + "update_plan", + "update_plan" + ], + [ + "update_plan" + ], + [ + "answer" + ] + ], + "citeable_context_ids": [ + "pqac-00000000", + "pqac-00000001", + "pqac-00000002", + "pqac-00000003", + "pqac-00000004", + "pqac-00000005", + "pqac-00000006", + "pqac-00000007", + "pqac-00000008", + "pqac-00000009", + "pqac-00000010", + "pqac-00000011", + "pqac-00000012", + "pqac-00000013", + "pqac-00000014", + "pqac-00000015", + "pqac-00000016", + "pqac-00000017" + ], + "artifacts": { + "artifact-00": "| Area | Closest published work(s) | What is automated | Objective/criterion | What is explicitly not automated | Gap assessment |\n|---|---|---|---|---|---|\n| 1. Optimizer-driven tuning of EBSD pattern-processing or indexing parameters | **Closest demonstrated EBSD parameter sweep:** Polonsky et al. 2019 parameterized DREAM.3D reconstruction variables and swept grain-segmentation misorientation tolerance over 0.1\u201310\u00b0 in 0.1\u00b0 increments, generating 100 reconstructions; **nearby but different:** Zhu et al. 2022 is only available here as metadata for \u201cglobal optimization algorithms\u201d applied to orientation/pattern-center/deformation extraction, so it should be treated as adjacent prior art, not verified from text in this record; **nearby non-tuning automation:** Kaufmann et al. 2020 CNN phase mapping automates phase recognition from patterns rather than tuning EBSD processing settings (pqac-00000014, pqac-00000013, pqac-00000004, pqac-00000005) | Automated batch execution / parameterization of reconstruction pipelines in BisQue+DREAM.3D; CNN-based direct phase classification from patterns in Kaufmann 2020 (pqac-00000014, pqac-00000004) | Polonsky used reconstruction-level metrics: fraction of \u201cgood\u201d CI-masked voxels retained and total grain count, selecting an elbow/balance point at 2.0\u00b0; Kaufmann targets phase-map fidelity qualitatively against known microstructure/EDS, not optimizer search (pqac-00000013, pqac-00000005) | No evidence here of Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, RL, random/grid search over **pattern** processing settings such as background correction, binning, neighbor averaging, Hough band count, rho-theta smoothing, or dictionary settings. Polonsky optimizes segmentation/reconstruction variables after indexing, not indexing settings themselves. Kaufmann does not tune EBSD parameters at all (pqac-00000013, pqac-00000005) | **Gap largely open.** Closest evidence is reconstruction-pipeline parameterization, not closed-loop optimizer-driven tuning of pattern-processing/indexing knobs. |\n| 2. Feedback loops that re-index stored raw patterns using CI/IQ/MAD/NCC to choose next parameter set | **Closest proposal:** Shen et al. 2019 proposes combining fast CNN predictions with a confidence measure (quaternion norm) so that high-confidence points get neighborhood dictionary search and low-confidence points get full dictionary search; **related raw-pattern availability, not loop:** Stinville et al. 2022 notes raw .up2 patterns can be opened in EMsoft DI / EMSphInx, enabling reindexing in principle (pqac-00000017, pqac-00000015, pqac-00000002, pqac-00000003) | Shen automates confidence estimation for each location and proposes routing subsequent dictionary matching effort based on confidence; raw pattern archival in Stinville enables offline reindexing workflows (pqac-00000017, pqac-00000003) | Quaternion norm as confidence proxy in Shen; no demonstrated CI/IQ/MAD/NCC-driven next-parameter selection loop in the gathered evidence (pqac-00000017) | Shen does **not** demonstrate an executed closed loop that iteratively changes indexing/processing parameters and re-indexes stored raw patterns to maximize CI/IQ/MAD/NCC/NCC-like objectives; it is a proposed hybrid search allocation strategy. Stinville provides data access, not autonomous optimization (pqac-00000017, pqac-00000003) | **Gap open.** Proposal-level hybrid confidence routing exists, but not a demonstrated feedback optimizer that selects the next EBSD parameter set from reindexed raw patterns. |\n| 3. Optimization against grain-level or microstructure-level outcomes | **Closest demonstrated:** Polonsky et al. 2019 explicitly optimizes a grain-segmentation parameter against microstructure-level outcomes (grain count and retained good-data fraction); Stan\u011bk et al. 2020 compares 2D/3D segmentation sensitivity to misorientation threshold and shows strong threshold dependence, but not autonomous optimization (pqac-00000013, pqac-00000012, pqac-00000010, pqac-00000011) | Automated parallel reconstructions and post-processing counts in Polonsky; analytical comparison of threshold effects on segmentation in Stan\u011bk (pqac-00000013, pqac-00000010) | Grain count stability / avoidance of grain merging plus retained CI-masked voxel fraction in Polonsky; sensitivity of number of grains and segmentation discrepancies vs threshold in Stan\u011bk (pqac-00000013, pqac-00000011) | No demonstrated multi-objective or constrained optimization over broader grain-level cleanup outcomes such as speckle/wild-grain fraction, grain-boundary integrity, fraction of pixels altered by cleanup, or Pareto balancing across multiple EBSD post-processing stages. Polonsky is a one-parameter sweep with manually interpreted elbow criterion (pqac-00000013, pqac-00000012) | **Partially occupied, but narrow.** Grain-level optimization exists for segmentation tolerance; broader autonomous multi-objective EBSD post-processing remains a gap. |\n| 4. Hybrid Hough-first / dictionary-on-residue and automated escalation criteria | **Closest published work:** Shen et al. 2019 explicitly proposes a hybrid strategy: CNN preliminary prediction plus confidence, then pattern matching only in a neighborhood for high-confidence points and full dictionary search for low-confidence points; this is the clearest automated escalation criterion in the gathered evidence. Sparks et al. 2024 and Winiarski et al. 2021 automate registration/distortion correction workflows but not Hough\u2192dictionary escalation (pqac-00000017, pqac-00000000, pqac-00000008) | Automated confidence scoring and proposed routing of dictionary-search scope in Shen; automated image registration/distortion correction in Sparks/Winiarski (pqac-00000017, pqac-00000000, pqac-00000008) | Criterion in Shen is confidence from quaternion norm: high confidence \u2192 local neighborhood dictionary search; low confidence \u2192 full dictionary search (pqac-00000017) | No evidence in the gathered text of a **demonstrated** Hough-first then dictionary-on-low-quality-residue loop with automated escalation based on CI/IQ/MAD/NCC thresholds. Shen\u2019s routing is proposed around CNN+dictionary, not Hough+dictionary, and not shown as an executed autonomous residue-reindexing pipeline (pqac-00000017) | **Mostly gap.** There is a close proposal for automated escalation, but not a demonstrated Hough-residue-to-dictionary autonomous workflow. |\n| 5. Autonomous/self-driving EBSD post-processing or characterization workflows, including links back to SEM acquisition recommendations | **Closest workflow automation:** Sparks et al. 2024 automates sample/image registration steps, pattern-center fitting for every section, and alignment routines; Winiarski et al. 2021 automates SE\u2013EBSD distortion correction and argues more automated procedures are needed; Burnett et al. 2016 demonstrates a fully automated serial-sectioning routine with tracking fiducials, but acquisition settings were experimentally chosen; Kaufmann et al. 2020 automates phase mapping from patterns (pqac-00000001, pqac-00000000, pqac-00000008, pqac-00000009, pqac-00000006, pqac-00000007, pqac-00000004) | Automated registration/warping, per-layer EBSD-to-BSE alignment, CI-threshold masks, pattern-center fitting per section, and automated serial sectioning with fiducial tracking; CNN inference for phase maps (pqac-00000000, pqac-00000001, pqac-00000006, pqac-00000004) | Workflow goals are geometric consistency, low inter-section misorientation, and corrected distortions; Burnett assesses equivalent quality via EBSD pattern quality/band contrast and indexing rate after experimentally selected settings (pqac-00000001, pqac-00000008, pqac-00000007) | No evidence here of a human-out-of-the-loop EBSD post-processing system that autonomously tunes post-processing/indexing parameters end-to-end, nor of a loop that feeds post-processing outcomes back to SEM acquisition parameters such as beam current, dwell time, or step size. Burnett explicitly determines acceptable milling current experimentally; Winiarski says more automated procedures still need to be developed (pqac-00000007, pqac-00000008) | **Gap open.** Sub-workflow automation exists, but not a self-driving closed-loop EBSD characterization system that recommends or controls acquisition parameters from post-processing outcomes. |\n\n\n*Table: This table maps the closest published prior art found for the five requested EBSD automation areas, distinguishing demonstrated methods from proposals and from absent closed loops. It highlights that most evidence clusters around workflow automation and grain-segmentation parameter sweeps, not autonomous closed-loop tuning of indexing/processing parameters.*" + }, + "candidates": { + "call_TMwBuCDwZNGWGVmRuRrINX4G": [], + "call_9At2TUTAjnJ23z8NGaLTDASo": [ + { + "embedding": null, + "docname": "panchi2023bayesianoptimizationfor", + "dockey": "15dd6fed0aaf24ad", + "citation": "Navid Panchi. Bayesian optimization for diffraction profile modeling. Proceedings of the Institute for Multiscale Simulation, Jan 2023. URL: https://doi.org/10.25593/open-fau-1463, doi:10.25593/open-fau-1463.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "panchi2023bayesianoptimizationfor", + "bibtex": "@article{panchi2023bayesianoptimizationfor,\n author = \"Panchi, Navid\",\n title = \"Bayesian Optimization for Diffraction Profile Modeling\",\n year = \"2023\",\n journal = \"Proceedings of the Institute for Multiscale Simulation\",\n month = \"Jan\",\n doi = \"10.25593/open-fau-1463\",\n url = \"https://doi.org/10.25593/open-fau-1463\",\n publisher = \"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)\"\n}\n", + "authors": [ + "Navid Panchi" + ], + "publication_date": "2023-01-20T00:00:00Z", + "year": 2023, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Proceedings of the Institute for Multiscale Simulation", + "publisher": "Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)", + "url": "https://open.fau.de/handle/openfau/33541", + "title": "Bayesian Optimization for Diffraction Profile Modeling", + "citation_count": 0, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.25593/open-fau-1463", + "doi_url": "https://doi.org/10.25593/open-fau-1463", + "doc_id": "15dd6fed0aaf24ad", + "file_location": null, + "license": null, + "pdf_url": null, + "other": { + "client_source": [ + "paper_search", + "datacite", + "datacite", + "datacite", + "datacite" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Navid Panchi. Bayesian optimization for diffraction profile modeling. Proceedings of the Institute for Multiscale Simulation, Jan 2023. URL: https://doi.org/10.25593/open-fau-1463, doi:10.25593/open-fau-1463. This article has 0 citations." + }, + { + "embedding": null, + "docname": "frydrych2021materialsinformaticsfor", + "dockey": "dc95f8c662e906cd", + "citation": "Karol Frydrych, Kamran Karimi, Michal Pecelerowicz, Rene Alvarez, Francesco Javier Dominguez-Guti\u00e9rrez, Fabrizio Rovaris, and Stefanos Papanikolaou. Materials informatics for mechanical deformation: a review of applications and challenges. Materials, 14:5764, Oct 2021. URL: https://doi.org/10.3390/ma14195764, doi:10.3390/ma14195764.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "frydrych2021materialsinformaticsfor", + "bibtex": "@article{frydrych2021materialsinformaticsfor,\n author = \"Frydrych, Karol and Karimi, Kamran and Pecelerowicz, Michal and Alvarez, Rene and Dominguez-Guti\u00e9rrez, Francesco Javier and Rovaris, Fabrizio and Papanikolaou, Stefanos\",\n title = \"Materials Informatics for Mechanical Deformation: A Review of Applications and Challenges\",\n year = \"2021\",\n journal = \"Materials\",\n volume = \"14\",\n pages = \"5764\",\n month = \"Oct\",\n doi = \"10.3390/ma14195764\",\n url = \"https://doi.org/10.3390/ma14195764\",\n publisher = \"MDPI AG\",\n issue = \"19\",\n issn = \"1996-1944\"\n}\n", + "authors": [ + "Karol Frydrych", + "Kamran Karimi", + "Michal Pecelerowicz", + "Rene Alvarez", + "Francesco Javier Dominguez-Guti\u00e9rrez", + "Fabrizio Rovaris", + "Stefanos Papanikolaou" + ], + "publication_date": "2021-10-02T00:00:00Z", + "year": 2021, + "volume": "14", + "issue": "19", + "issn": "1996-1944", + "pages": "5764", + "journal": "Materials", + "publisher": "MDPI AG", + "url": "https://doi.org/10.3390/ma14195764", + "title": "Materials Informatics for Mechanical Deformation: A Review of Applications and Challenges", + "citation_count": 80, + "bibtex_type": "misc", + "source_quality": -1, + "is_retracted": null, + "doi": "10.3390/ma14195764", + "doi_url": "https://doi.org/10.3390/ma14195764", + "doc_id": "dc95f8c662e906cd", + "file_location": null, + "license": null, + "pdf_url": "https://www.mdpi.com/1996-1944/14/19/5764/pdf?version=1633167246", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Karol Frydrych, Kamran Karimi, Michal Pecelerowicz, Rene Alvarez, Francesco Javier Dominguez-Guti\u00e9rrez, Fabrizio Rovaris, and Stefanos Papanikolaou. Materials informatics for mechanical deformation: a review of applications and challenges. Materials, 14:5764, Oct 2021. URL: https://doi.org/10.3390/ma14195764, doi:10.3390/ma14195764. This article has 80 citations." + }, + { + "embedding": null, + "docname": "stinville2022multimodaldatasetof", + "dockey": "e97010568afd0695", + "citation": "J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stinville2022multimodaldatasetof", + "bibtex": "@article{stinville2022multimodaldatasetof,\n author = \"Stinville, J. C. and Hestroffer, J. M. and Charpagne, M. A. and Polonsky, A. T. and Echlin, M. P. and Torbet, C. J. and Valle, V. and Nygren, K. E. and Miller, M. P. and Klaas, O. and Loghin, A. and Beyerlein, I. J. and Pollock, T. M.\",\n title = \"Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields\",\n year = \"2022\",\n journal = \"Scientific Data\",\n volume = \"9\",\n month = \"Aug\",\n doi = \"10.1038/s41597-022-01525-w\",\n url = \"https://doi.org/10.1038/s41597-022-01525-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2052-4463\"\n}\n", + "authors": [ + "J. C. Stinville", + "J. M. Hestroffer", + "M. A. Charpagne", + "A. T. Polonsky", + "M. P. Echlin", + "C. J. Torbet", + "V. Valle", + "K. E. Nygren", + "M. P. Miller", + "O. Klaas", + "A. Loghin", + "I. J. Beyerlein", + "T. M. Pollock" + ], + "publication_date": "2022-08-01T00:00:00Z", + "year": 2022, + "volume": "9", + "issue": "1", + "issn": "2052-4463", + "pages": null, + "journal": "Scientific Data", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41597-022-01525-w", + "title": "Multi-modal Dataset of a Polycrystalline Metallic Material: 3D Microstructure and Deformation Fields", + "citation_count": 36, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41597-022-01525-w", + "doi_url": "https://doi.org/10.1038/s41597-022-01525-w", + "doc_id": "e97010568afd0695", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41597-022-01525-w.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "J. C. Stinville, J. M. Hestroffer, M. A. Charpagne, A. T. Polonsky, M. P. Echlin, C. J. Torbet, V. Valle, K. E. Nygren, M. P. Miller, O. Klaas, A. Loghin, I. J. Beyerlein, and T. M. Pollock. Multi-modal dataset of a polycrystalline metallic material: 3d microstructure and deformation fields. Scientific Data, Aug 2022. URL: https://doi.org/10.1038/s41597-022-01525-w, doi:10.1038/s41597-022-01525-w. This article has 36 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "batra2021emergingmaterialsintelligence", + "dockey": "ab748ce1292815c0", + "citation": "Rohit Batra, Le Song, and Rampi Ramprasad. Emerging materials intelligence ecosystems propelled by machine learning. Nature Reviews Materials, 6:655-678, Nov 2021. URL: https://doi.org/10.1038/s41578-020-00255-y, doi:10.1038/s41578-020-00255-y.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "batra2021emergingmaterialsintelligence", + "bibtex": "@article{batra2021emergingmaterialsintelligence,\n author = \"Batra, Rohit and Song, Le and Ramprasad, Rampi\",\n title = \"Emerging materials intelligence ecosystems propelled by machine learning\",\n year = \"2021\",\n journal = \"Nature Reviews Materials\",\n volume = \"6\",\n pages = \"655-678\",\n month = \"Nov\",\n doi = \"10.1038/s41578-020-00255-y\",\n url = \"https://doi.org/10.1038/s41578-020-00255-y\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"8\",\n issn = \"2058-8437\"\n}\n", + "authors": [ + "Rohit Batra", + "Le Song", + "Rampi Ramprasad" + ], + "publication_date": "2020-11-09T00:00:00Z", + "year": 2021, + "volume": "6", + "issue": "8", + "issn": "2058-8437", + "pages": "655-678", + "journal": "Nature Reviews Materials", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41578-020-00255-y", + "title": "Emerging materials intelligence ecosystems propelled by machine learning", + "citation_count": 513, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.1038/s41578-020-00255-y", + "doi_url": "https://doi.org/10.1038/s41578-020-00255-y", + "doc_id": "ab748ce1292815c0", + "file_location": null, + "license": null, + "pdf_url": "https://www.osti.gov/servlets/purl/1864296", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Rohit Batra, Le Song, and Rampi Ramprasad. Emerging materials intelligence ecosystems propelled by machine learning. Nature Reviews Materials, 6:655-678, Nov 2021. URL: https://doi.org/10.1038/s41578-020-00255-y, doi:10.1038/s41578-020-00255-y. This article has 513 citations." + } + ], + "call_5aBBEXJupU0KmlHg00b79lgu": [ + { + "embedding": null, + "docname": "amini2024sustainabilityinspiredfabrication", + "dockey": "911096d6a6cb0eae", + "citation": "Shahram Amini, Hongbin Choi, Wesley Seche, Alexander Blagojevic, Nicholas May, Benjamin M. Lefler, Skyler L. Davis, Sahar Elyahoodayan, Pouya Tavousi, Steven J. May, Gregory A. Caputo, Terry C. Lowe, Jeffrey Hettinger, and Sina Shahbazmohamadi. Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring. Microsystems & nanoengineering, 10 1:125, Sep 2024. URL: https://doi.org/10.1038/s41378-024-00754-w, doi:10.1038/s41378-024-00754-w.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "amini2024sustainabilityinspiredfabrication", + "bibtex": "@article{amini2024sustainabilityinspiredfabrication,\n author = \"Amini, Shahram and Choi, Hongbin and Seche, Wesley and Blagojevic, Alexander and May, Nicholas and Lefler, Benjamin M. and Davis, Skyler L. and Elyahoodayan, Sahar and Tavousi, Pouya and May, Steven J. and Caputo, Gregory A. and Lowe, Terry C. and Hettinger, Jeffrey and Shahbazmohamadi, Sina\",\n title = \"Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring.\",\n year = \"2024\",\n journal = \"Microsystems \\& nanoengineering\",\n volume = \"10 1\",\n pages = \"125\",\n month = \"Sep\",\n doi = \"10.1038/s41378-024-00754-w\",\n url = \"https://doi.org/10.1038/s41378-024-00754-w\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2055-7434\"\n}\n", + "authors": [ + "Shahram Amini", + "Hongbin Choi", + "Wesley Seche", + "Alexander Blagojevic", + "Nicholas May", + "Benjamin M. Lefler", + "Skyler L. Davis", + "Sahar Elyahoodayan", + "Pouya Tavousi", + "Steven J. May", + "Gregory A. Caputo", + "Terry C. Lowe", + "Jeffrey Hettinger", + "Sina Shahbazmohamadi" + ], + "publication_date": "2024-09-09T00:00:00Z", + "year": 2024, + "volume": "10 1", + "issue": "1", + "issn": "2055-7434", + "pages": "125", + "journal": "Microsystems & nanoengineering", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41378-024-00754-w", + "title": "Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring.", + "citation_count": 8, + "bibtex_type": "article", + "source_quality": 2, + "is_retracted": null, + "doi": "10.1038/s41378-024-00754-w", + "doi_url": "https://doi.org/10.1038/s41378-024-00754-w", + "doc_id": "911096d6a6cb0eae", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41378-024-00754-w.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Shahram Amini, Hongbin Choi, Wesley Seche, Alexander Blagojevic, Nicholas May, Benjamin M. Lefler, Skyler L. Davis, Sahar Elyahoodayan, Pouya Tavousi, Steven J. May, Gregory A. Caputo, Terry C. Lowe, Jeffrey Hettinger, and Sina Shahbazmohamadi. Sustainability inspired fabrication of next generation neurostimulation and cardiac rhythm management electrodes via reactive hierarchical surface restructuring. Microsystems & nanoengineering, 10 1:125, Sep 2024. URL: https://doi.org/10.1038/s41378-024-00754-w, doi:10.1038/s41378-024-00754-w. This article has 8 citations and is from a domain leading peer-reviewed journal." + } + ], + "call_acsPy2saVbrPRcVvQrHXhwqx": [], + "call_SvSRR7s0yK3PpnhQQ0Vw5Xl4": [ + { + "embedding": null, + "docname": "choudhary2022recentadvancesand", + "dockey": "50bfe20b4abf8c33", + "citation": "Kamal Choudhary, Brian DeCost, Chi Chen, Anubhav Jain, Francesca Tavazza, Ryan Cohn, Cheol Woo Park, Alok Choudhary, Ankit Agrawal, Simon J. L. Billinge, Elizabeth Holm, Shyue Ping Ong, and Chris Wolverton. Recent advances and applications of deep learning methods in materials science. npj Computational Materials, 8:1-26, Apr 2022. URL: https://doi.org/10.1038/s41524-022-00734-6, doi:10.1038/s41524-022-00734-6.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "choudhary2022recentadvancesand", + "bibtex": "@article{choudhary2022recentadvancesand,\n author = \"Choudhary, Kamal and DeCost, Brian and Chen, Chi and Jain, Anubhav and Tavazza, Francesca and Cohn, Ryan and Park, Cheol Woo and Choudhary, Alok and Agrawal, Ankit and Billinge, Simon J. L. and Holm, Elizabeth and Ong, Shyue Ping and Wolverton, Chris\",\n title = \"Recent advances and applications of deep learning methods in materials science\",\n year = \"2022\",\n journal = \"npj Computational Materials\",\n volume = \"8\",\n month = \"Apr\",\n doi = \"10.1038/s41524-022-00734-6\",\n url = \"https://doi.org/10.1038/s41524-022-00734-6\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\",\n pages = \"1-26\"\n}\n", + "authors": [ + "Kamal Choudhary", + "Brian DeCost", + "Chi Chen", + "Anubhav Jain", + "Francesca Tavazza", + "Ryan Cohn", + "Cheol Woo Park", + "Alok Choudhary", + "Ankit Agrawal", + "Simon J. L. Billinge", + "Elizabeth Holm", + "Shyue Ping Ong", + "Chris Wolverton" + ], + "publication_date": "2022-04-05T00:00:00Z", + "year": 2022, + "volume": "8", + "issue": "1", + "issn": "2057-3960", + "pages": "1-26", + "journal": "npj Computational Materials", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41524-022-00734-6", + "title": "Recent advances and applications of deep learning methods in materials science", + "citation_count": 1030, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41524-022-00734-6", + "doi_url": "https://doi.org/10.1038/s41524-022-00734-6", + "doc_id": "50bfe20b4abf8c33", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41524-022-00734-6.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "Kamal Choudhary, Brian DeCost, Chi Chen, Anubhav Jain, Francesca Tavazza, Ryan Cohn, Cheol Woo Park, Alok Choudhary, Ankit Agrawal, Simon J. L. Billinge, Elizabeth Holm, Shyue Ping Ong, and Chris Wolverton. Recent advances and applications of deep learning methods in materials science. npj Computational Materials, 8:1-26, Apr 2022. URL: https://doi.org/10.1038/s41524-022-00734-6, doi:10.1038/s41524-022-00734-6. This article has 1030 citations and is from a peer-reviewed journal." + } + ], + "call_yBdeyyzRACDmvHLG4V5YZoxC": [ + { + "embedding": null, + "docname": "sparks20243dreconstructionof", + "dockey": "be99dadb0f50f040", + "citation": "Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "sparks20243dreconstructionof", + "bibtex": "@article{sparks20243dreconstructionof,\n author = \"Sparks, Gregory and Mason, Simon A. and Chapman, Michael G. and Park, Jun-Sang and Sharma, Hemant and Kenesei, Peter and Niezgoda, Stephen R. and Mills, Michael J. and Uchic, Michael D. and Shade, Paul A. and Obstalecki, Mark\",\n title = \"3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jul\",\n doi = \"10.1007/s40192-024-00370-6\",\n url = \"https://doi.org/10.1007/s40192-024-00370-6\",\n volume = \"13\",\n pages = \"773-803\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"3\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Gregory Sparks", + "Simon A. Mason", + "Michael G. Chapman", + "Jun-Sang Park", + "Hemant Sharma", + "Peter Kenesei", + "Stephen R. Niezgoda", + "Michael J. Mills", + "Michael D. Uchic", + "Paul A. Shade", + "Mark Obstalecki" + ], + "publication_date": "2024-07-24T00:00:00Z", + "year": 2024, + "volume": "13", + "issue": "3", + "issn": "2193-9764", + "pages": "773-803", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-024-00370-6", + "title": "3D Reconstruction of a High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning with High-Precision EBSD and Surface Profilometry", + "citation_count": 11, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-024-00370-6", + "doi_url": "https://doi.org/10.1007/s40192-024-00370-6", + "doc_id": "be99dadb0f50f040", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-024-00370-6.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Gregory Sparks, Simon A. Mason, Michael G. Chapman, Jun-Sang Park, Hemant Sharma, Peter Kenesei, Stephen R. Niezgoda, Michael J. Mills, Michael D. Uchic, Paul A. Shade, and Mark Obstalecki. 3d reconstruction of a high-energy diffraction microscopy sample using multi-modal serial sectioning with high-precision ebsd and surface profilometry. Integrating Materials and Manufacturing Innovation, 13:773-803, Jul 2024. URL: https://doi.org/10.1007/s40192-024-00370-6, doi:10.1007/s40192-024-00370-6. This article has 11 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "chen2025recentprogressof", + "dockey": "5811c2803a65a250", + "citation": "Bingbing Chen, Dongfeng Li, Peter Davies, Richard Johnston, Xiangyun Ge, and Chenfeng Li. Recent progress of digital reconstruction in polycrystalline materials. Archives of Computational Methods in Engineering, Mar 2025. URL: https://doi.org/10.1007/s11831-025-10245-4, doi:10.1007/s11831-025-10245-4.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "chen2025recentprogressof", + "bibtex": "@article{chen2025recentprogressof,\n author = \"Chen, Bingbing and Li, Dongfeng and Davies, Peter and Johnston, Richard and Ge, Xiangyun and Li, Chenfeng\",\n title = \"Recent Progress of Digital Reconstruction in Polycrystalline Materials\",\n year = \"2025\",\n journal = \"Archives of Computational Methods in Engineering\",\n month = \"Mar\",\n doi = \"10.1007/s11831-025-10245-4\",\n url = \"https://doi.org/10.1007/s11831-025-10245-4\"\n}\n", + "authors": [ + "Bingbing Chen", + "Dongfeng Li", + "Peter Davies", + "Richard Johnston", + "Xiangyun Ge", + "Chenfeng Li" + ], + "publication_date": "2025-03-01T00:00:00Z", + "year": 2025, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Archives of Computational Methods in Engineering", + "publisher": null, + "url": "https://link.springer.com/content/pdf/10.1007/s11831-025-10245-4.pdf", + "title": "Recent Progress of Digital Reconstruction in Polycrystalline Materials", + "citation_count": 13, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s11831-025-10245-4", + "doi_url": "https://doi.org/10.1007/s11831-025-10245-4", + "doc_id": "5811c2803a65a250", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s11831-025-10245-4.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Bingbing Chen, Dongfeng Li, Peter Davies, Richard Johnston, Xiangyun Ge, and Chenfeng Li. Recent progress of digital reconstruction in polycrystalline materials. Archives of Computational Methods in Engineering, Mar 2025. URL: https://doi.org/10.1007/s11831-025-10245-4, doi:10.1007/s11831-025-10245-4. This article has 13 citations and is from a peer-reviewed journal." + } + ], + "call_4uZ0w5ewQsn4Ra9JRMzH48rh": [], + "call_WJzVeGxOMudv7uXJmeYdSe3F": [ + { + "embedding": null, + "docname": "US12517250B2", + "dockey": "5638a85b49fd6858", + "citation": "Nicholas A. Kotov. Multidimensional materials sensing systems and methods. Patent (WO,EP,US,CN,JP,KR), 2026.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "US12517250B2", + "bibtex": "@article{US12517250B2,\n author = \"Kotov, Nicholas A.\",\n title = \"Multidimensional materials sensing systems and methods\",\n year = \"2026\",\n journal = \"Patent (WO,EP,US,CN,JP,KR)\"\n}\n", + "authors": [ + "Nicholas A. Kotov" + ], + "publication_date": null, + "year": 2026, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Patent (WO,EP,US,CN,JP,KR)", + "publisher": null, + "url": "https://patents.google.com/patent/US12517250B2/en", + "title": "Multidimensional materials sensing systems and methods", + "citation_count": null, + "bibtex_type": null, + "source_quality": null, + "is_retracted": null, + "doi": null, + "doi_url": null, + "doc_id": "5638a85b49fd6858", + "file_location": null, + "license": null, + "pdf_url": null, + "other": { + "client_source": [ + "patent_search" + ], + "bibtex_source": [ + "self_generated" + ] + }, + "formatted_citation": "Nicholas A. Kotov. Multidimensional materials sensing systems and methods. Patent (WO,EP,US,CN,JP,KR), 2026." + }, + { + "embedding": null, + "docname": "US20220276352A1", + "dockey": "3d314f0ddfb6f82d", + "citation": "Ricardo Ferreira. Optical package for a lidar sensor system and lidar sensor system technical \u2026. Patent (WO,EP,US,CN,CA,DE), 2022.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "US20220276352A1", + "bibtex": "@article{US20220276352A1,\n author = \"Ferreira, Ricardo\",\n title = \"Optical package for a lidar sensor system and lidar sensor system technical \u2026\",\n year = \"2022\",\n journal = \"Patent (WO,EP,US,CN,CA,DE)\"\n}\n", + "authors": [ + "Ricardo Ferreira" + ], + "publication_date": null, + "year": 2022, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Patent (WO,EP,US,CN,CA,DE)", + "publisher": null, + "url": "https://patents.google.com/patent/US20220276352A1/en", + "title": "Optical package for a lidar sensor system and lidar sensor system technical \u2026", + "citation_count": null, + "bibtex_type": null, + "source_quality": null, + "is_retracted": null, + "doi": null, + "doi_url": null, + "doc_id": "3d314f0ddfb6f82d", + "file_location": null, + "license": null, + "pdf_url": "https://patentimages.storage.googleapis.com/69/bd/19/03e2eac191be65/US20220276352A1.pdf", + "other": { + "client_source": [ + "patent_search" + ], + "bibtex_source": [ + "self_generated" + ] + }, + "formatted_citation": "Ricardo Ferreira. Optical package for a lidar sensor system and lidar sensor system technical \u2026. Patent (WO,EP,US,CN,CA,DE), 2022." + }, + { + "embedding": null, + "docname": "US11112503B2", + "dockey": "bb332753ed5711a5", + "citation": "Dale G. Fried. Methods and apparatus for three-dimensional (3d) imaging. Patent (WO,EP,US,CA), 2021.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "US11112503B2", + "bibtex": "@article{US11112503B2,\n author = \"Fried, Dale G.\",\n title = \"Methods and apparatus for three-dimensional (3D) imaging\",\n year = \"2021\",\n journal = \"Patent (WO,EP,US,CA)\"\n}\n", + "authors": [ + "Dale G. Fried" + ], + "publication_date": null, + "year": 2021, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Patent (WO,EP,US,CA)", + "publisher": null, + "url": "https://patents.google.com/patent/US11112503B2/en", + "title": "Methods and apparatus for three-dimensional (3D) imaging", + "citation_count": null, + "bibtex_type": null, + "source_quality": null, + "is_retracted": null, + "doi": null, + "doi_url": null, + "doc_id": "bb332753ed5711a5", + "file_location": null, + "license": null, + "pdf_url": "https://patentimages.storage.googleapis.com/4d/4e/e7/01acace611ddde/US11112503.pdf", + "other": { + "client_source": [ + "patent_search" + ], + "bibtex_source": [ + "self_generated" + ] + }, + "formatted_citation": "Dale G. Fried. Methods and apparatus for three-dimensional (3d) imaging. Patent (WO,EP,US,CA), 2021." + }, + { + "embedding": null, + "docname": "US10046229B2", + "dockey": "cffd382dd06367eb", + "citation": "Bao Tran. Smart device. Patent (US), 2018.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "US10046229B2", + "bibtex": "@article{US10046229B2,\n author = \"Tran, Bao\",\n title = \"Smart device\",\n year = \"2018\",\n journal = \"Patent (US)\"\n}\n", + "authors": [ + "Bao Tran" + ], + "publication_date": null, + "year": 2018, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Patent (US)", + "publisher": null, + "url": "https://patents.google.com/patent/US10046229B2/en", + "title": "Smart device", + "citation_count": null, + "bibtex_type": null, + "source_quality": null, + "is_retracted": null, + "doi": null, + "doi_url": null, + "doc_id": "cffd382dd06367eb", + "file_location": null, + "license": null, + "pdf_url": "https://patentimages.storage.googleapis.com/81/f8/d8/1bcade3ce5a938/US10046229.pdf", + "other": { + "client_source": [ + "patent_search" + ], + "bibtex_source": [ + "self_generated" + ] + }, + "formatted_citation": "Bao Tran. Smart device. Patent (US), 2018." + }, + { + "embedding": null, + "docname": "US11320588B1", + "dockey": "dab2a7d5dd62b087", + "citation": "Mohammad A. Mazed. Super system on chip. Patent (US), 2022.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "US11320588B1", + "bibtex": "@article{US11320588B1,\n author = \"Mazed, Mohammad A.\",\n title = \"Super system on chip\",\n year = \"2022\",\n journal = \"Patent (US)\"\n}\n", + "authors": [ + "Mohammad A. Mazed" + ], + "publication_date": null, + "year": 2022, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Patent (US)", + "publisher": null, + "url": "https://patents.google.com/patent/US11320588B1/en", + "title": "Super system on chip", + "citation_count": null, + "bibtex_type": null, + "source_quality": null, + "is_retracted": null, + "doi": null, + "doi_url": null, + "doc_id": "dab2a7d5dd62b087", + "file_location": null, + "license": null, + "pdf_url": "https://patentimages.storage.googleapis.com/91/df/cd/56a746922cebc0/US11320588.pdf", + "other": { + "client_source": [ + "patent_search" + ], + "bibtex_source": [ + "self_generated" + ] + }, + "formatted_citation": "Mohammad A. Mazed. Super system on chip. Patent (US), 2022." + }, + { + "embedding": null, + "docname": "US11650190B2", + "dockey": "1bc7cbfd8245358e", + "citation": "Nicholas Tufillaro. Hyperspectral sensing system and methods. Patent (US), 2023.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "US11650190B2", + "bibtex": "@article{US11650190B2,\n author = \"Tufillaro, Nicholas\",\n title = \"Hyperspectral sensing system and methods\",\n year = \"2023\",\n journal = \"Patent (US)\"\n}\n", + "authors": [ + "Nicholas Tufillaro" + ], + "publication_date": null, + "year": 2023, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Patent (US)", + "publisher": null, + "url": "https://patents.google.com/patent/US11650190B2/en", + "title": "Hyperspectral sensing system and methods", + "citation_count": null, + "bibtex_type": null, + "source_quality": null, + "is_retracted": null, + "doi": null, + "doi_url": null, + "doc_id": "1bc7cbfd8245358e", + "file_location": null, + "license": null, + "pdf_url": "https://patentimages.storage.googleapis.com/c7/a3/c1/cfc8309154c3be/US11650190.pdf", + "other": { + "client_source": [ + "patent_search" + ], + "bibtex_source": [ + "self_generated" + ] + }, + "formatted_citation": "Nicholas Tufillaro. Hyperspectral sensing system and methods. Patent (US), 2023." + }, + { + "embedding": null, + "docname": "US12380059B2", + "dockey": "a94d4d56cdcdbccc", + "citation": "Mohammad A. Mazed. Super system on chip. Patent (US), 2025.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "US12380059B2", + "bibtex": "@article{US12380059B2,\n author = \"Mazed, Mohammad A.\",\n title = \"Super system on chip\",\n year = \"2025\",\n journal = \"Patent (US)\"\n}\n", + "authors": [ + "Mohammad A. Mazed" + ], + "publication_date": null, + "year": 2025, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Patent (US)", + "publisher": null, + "url": "https://patents.google.com/patent/US12380059B2/en", + "title": "Super system on chip", + "citation_count": null, + "bibtex_type": null, + "source_quality": null, + "is_retracted": null, + "doi": null, + "doi_url": null, + "doc_id": "a94d4d56cdcdbccc", + "file_location": null, + "license": null, + "pdf_url": null, + "other": { + "client_source": [ + "patent_search" + ], + "bibtex_source": [ + "self_generated" + ] + }, + "formatted_citation": "Mohammad A. Mazed. Super system on chip. Patent (US), 2025." + } + ], + "call_uh7IvP3yQ85eCVKwBxCbjRkg": [ + { + "embedding": null, + "docname": "timberlake2026effectsofsem", + "dockey": "17c61a4b08c6319e", + "citation": "D. M. Timberlake and E. J. Payton. Effects of sem distortions on ebsd grain size measurements. Metallography, Microstructure, and Analysis, 15:489-505, Feb 2026. URL: https://doi.org/10.1007/s13632-026-01317-3, doi:10.1007/s13632-026-01317-3.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "timberlake2026effectsofsem", + "bibtex": "@article{timberlake2026effectsofsem,\n author = \"Timberlake, D. M. and Payton, E. J.\",\n title = \"Effects of SEM Distortions on EBSD Grain Size Measurements\",\n year = \"2026\",\n journal = \"Metallography, Microstructure, and Analysis\",\n volume = \"15\",\n pages = \"489-505\",\n month = \"Feb\",\n doi = \"10.1007/s13632-026-01317-3\",\n url = \"https://doi.org/10.1007/s13632-026-01317-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2192-9262\"\n}\n", + "authors": [ + "D. M. Timberlake", + "E. J. Payton" + ], + "publication_date": "2026-02-23T00:00:00Z", + "year": 2026, + "volume": "15", + "issue": "2", + "issn": "2192-9262", + "pages": "489-505", + "journal": "Metallography, Microstructure, and Analysis", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s13632-026-01317-3", + "title": "Effects of SEM Distortions on EBSD Grain Size Measurements", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": -1, + "is_retracted": null, + "doi": "10.1007/s13632-026-01317-3", + "doi_url": "https://doi.org/10.1007/s13632-026-01317-3", + "doc_id": "17c61a4b08c6319e", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s13632-026-01317-3.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "D. M. Timberlake and E. J. Payton. Effects of sem distortions on ebsd grain size measurements. Metallography, Microstructure, and Analysis, 15:489-505, Feb 2026. URL: https://doi.org/10.1007/s13632-026-01317-3, doi:10.1007/s13632-026-01317-3. This article has 0 citations." + } + ], + "call_PGusrK3loi2SP0YX9X3dJxyb": [ + { + "embedding": null, + "docname": "ge2025highstrengthadditivelymanufacturable", + "dockey": "0cb5af61e034fb2f", + "citation": "Zhaoxuan Ge, Shaolou Wei, Zehua Liu, Tomasz Choma, Lukasz Zrodowski, Marc De Graef, A. John Hart, and S. Mohadeseh Taheri-Mousavi. High-strength additively manufacturable al-zr-er-ni alloys with high as-built ductility and thermal stability. Unknown journal, May 2025. URL: https://doi.org/10.21203/rs.3.rs-6558867/v1, doi:10.21203/rs.3.rs-6558867/v1.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "ge2025highstrengthadditivelymanufacturable", + "bibtex": "@article{ge2025highstrengthadditivelymanufacturable,\n author = \"Ge, Zhaoxuan and Wei, Shaolou and Liu, Zehua and Choma, Tomasz and Zrodowski, Lukasz and Graef, Marc De and Hart, A. John and Taheri-Mousavi, S. Mohadeseh\",\n title = \"High-strength additively manufacturable Al-Zr-Er-Ni alloys with high as-built ductility and thermal stability\",\n year = \"2025\",\n journal = \"Unknown journal\",\n month = \"May\",\n doi = \"10.21203/rs.3.rs-6558867/v1\",\n url = \"https://doi.org/10.21203/rs.3.rs-6558867/v1\",\n publisher = \"Springer Science and Business Media LLC\"\n}\n", + "authors": [ + "Zhaoxuan Ge", + "Shaolou Wei", + "Zehua Liu", + "Tomasz Choma", + "Lukasz Zrodowski", + "Marc De Graef", + "A. John Hart", + "S. Mohadeseh Taheri-Mousavi" + ], + "publication_date": "2025-05-07T00:00:00Z", + "year": 2025, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": null, + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.21203/rs.3.rs-6558867/v1", + "title": "High-strength additively manufacturable Al-Zr-Er-Ni alloys with high as-built ductility and thermal stability", + "citation_count": 3, + "bibtex_type": "misc", + "source_quality": null, + "is_retracted": null, + "doi": "10.21203/rs.3.rs-6558867/v1", + "doi_url": "https://doi.org/10.21203/rs.3.rs-6558867/v1", + "doc_id": "0cb5af61e034fb2f", + "file_location": null, + "license": null, + "pdf_url": "https://www.researchsquare.com/article/rs-6558867/latest.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "crossref", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Zhaoxuan Ge, Shaolou Wei, Zehua Liu, Tomasz Choma, Lukasz Zrodowski, Marc De Graef, A. John Hart, and S. Mohadeseh Taheri-Mousavi. High-strength additively manufacturable al-zr-er-ni alloys with high as-built ductility and thermal stability. Unknown journal, May 2025. URL: https://doi.org/10.21203/rs.3.rs-6558867/v1, doi:10.21203/rs.3.rs-6558867/v1." + }, + { + "embedding": null, + "docname": "kaufmann2020phasemappingin", + "dockey": "7fe5de81dfc311a6", + "citation": "Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "kaufmann2020phasemappingin", + "bibtex": "@article{kaufmann2020phasemappingin,\n author = \"Kaufmann, Kevin and Zhu, Chaoyi and Rosengarten, Alexander S. and Maryanovsky, Daniel and Wang, Haoren and Vecchio, Kenneth S.\",\n title = \"Phase Mapping in EBSD Using Convolutional Neural Networks\",\n year = \"2020\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"26\",\n pages = \"458-468\",\n month = \"May\",\n doi = \"10.1017/s1431927620001488\",\n url = \"https://doi.org/10.1017/s1431927620001488\",\n publisher = \"Oxford University Press (OUP)\",\n issue = \"3\",\n issn = \"1431-9276\"\n}\n", + "authors": [ + "Kevin Kaufmann", + "Chaoyi Zhu", + "Alexander S. Rosengarten", + "Daniel Maryanovsky", + "Haoren Wang", + "Kenneth S. Vecchio" + ], + "publication_date": "2020-05-11T00:00:00Z", + "year": 2020, + "volume": "26", + "issue": "3", + "issn": "1431-9276", + "pages": "458-468", + "journal": "Microscopy and Microanalysis", + "publisher": "Oxford University Press (OUP)", + "url": "https://www.cambridge.org/core/services/aop-cambridge-core/content/view/623282EB1DF6DB0559CEC8C74D30A74F/S1431927620001488a.pdf/div-class-title-phase-mapping-in-ebsd-using-convolutional-neural-networks-div.pdf", + "title": "Phase Mapping in EBSD Using Convolutional Neural Networks", + "citation_count": 34, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1017/s1431927620001488", + "doi_url": "https://doi.org/10.1017/s1431927620001488", + "doc_id": "7fe5de81dfc311a6", + "file_location": null, + "license": null, + "pdf_url": "https://academic.oup.com/mam/article-pdf/26/3/458/47996054/mam0458.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Kevin Kaufmann, Chaoyi Zhu, Alexander S. Rosengarten, Daniel Maryanovsky, Haoren Wang, and Kenneth S. Vecchio. Phase mapping in ebsd using convolutional neural networks. Microscopy and Microanalysis, 26:458-468, May 2020. URL: https://doi.org/10.1017/s1431927620001488, doi:10.1017/s1431927620001488. This article has 34 citations and is from a peer-reviewed journal." + } + ], + "call_KI8Kw21UgTxchIJAACgrtWy6": [ + { + "embedding": null, + "docname": "afsharnia2025artificialintelligenceand", + "dockey": "af490807c027607a", + "citation": "Habib Afsharnia and Javaid Butt. Artificial intelligence and machine learning in cold spray additive manufacturing: a systematic literature review. Journal of Manufacturing and Materials Processing, 9:334, Oct 2025. URL: https://doi.org/10.3390/jmmp9100334, doi:10.3390/jmmp9100334.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "afsharnia2025artificialintelligenceand", + "bibtex": "@article{afsharnia2025artificialintelligenceand,\n author = \"Afsharnia, Habib and Butt, Javaid\",\n title = \"Artificial Intelligence and Machine Learning in Cold Spray Additive Manufacturing: A Systematic Literature Review\",\n year = \"2025\",\n journal = \"Journal of Manufacturing and Materials Processing\",\n volume = \"9\",\n pages = \"334\",\n month = \"Oct\",\n doi = \"10.3390/jmmp9100334\",\n url = \"https://doi.org/10.3390/jmmp9100334\",\n publisher = \"MDPI AG\",\n issue = \"10\",\n issn = \"2504-4494\"\n}\n", + "authors": [ + "Habib Afsharnia", + "Javaid Butt" + ], + "publication_date": "2025-10-13T00:00:00Z", + "year": 2025, + "volume": "9", + "issue": "10", + "issn": "2504-4494", + "pages": "334", + "journal": "Journal of Manufacturing and Materials Processing", + "publisher": "MDPI AG", + "url": "https://doi.org/10.3390/jmmp9100334", + "title": "Artificial Intelligence and Machine Learning in Cold Spray Additive Manufacturing: A Systematic Literature Review", + "citation_count": 6, + "bibtex_type": "misc", + "source_quality": -1, + "is_retracted": null, + "doi": "10.3390/jmmp9100334", + "doi_url": "https://doi.org/10.3390/jmmp9100334", + "doc_id": "af490807c027607a", + "file_location": null, + "license": null, + "pdf_url": "https://www.mdpi.com/2504-4494/9/10/334/pdf?version=1760489299", + "other": { + "client_source": [ + "paper_search", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "Habib Afsharnia and Javaid Butt. Artificial intelligence and machine learning in cold spray additive manufacturing: a systematic literature review. Journal of Manufacturing and Materials Processing, 9:334, Oct 2025. URL: https://doi.org/10.3390/jmmp9100334, doi:10.3390/jmmp9100334. This article has 6 citations." + } + ], + "call_AFeXRuxf8pxFGq9DkB0yS3lq": [], + "call_rjeQSC8BesdXWTotR8LKnLxa": [ + { + "embedding": null, + "docname": "burnett2016largevolumeserial", + "dockey": "a2e8eb201167feb2", + "citation": "Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "burnett2016largevolumeserial", + "bibtex": "@article{burnett2016largevolumeserial,\n author = \"Burnett, Timothy L. and Burnett, Timothy L. and Kelley, R. and Winiarski, B. and Contreras, L. and Daly, Michael and Gholinia, A. and Burke, M. G. and Withers, P. J.\",\n title = \"Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.\",\n year = \"2016\",\n journal = \"Ultramicroscopy\",\n volume = \"161\",\n pages = \"119-129\",\n month = \"Feb\",\n doi = \"10.1016/j.ultramic.2015.11.001\",\n url = \"https://doi.org/10.1016/j.ultramic.2015.11.001\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n", + "authors": [ + "Timothy L. Burnett", + "Timothy L. Burnett", + "R. Kelley", + "B. Winiarski", + "L. Contreras", + "Michael Daly", + "A. Gholinia", + "M. G. Burke", + "P. J. Withers" + ], + "publication_date": "2016-02-01T00:00:00Z", + "year": 2016, + "volume": "161", + "issue": null, + "issn": "0304-3991", + "pages": "119-129", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2015.11.001", + "title": "Large volume serial section tomography by Xe Plasma FIB dual beam microscopy.", + "citation_count": 440, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2015.11.001", + "doi_url": "https://doi.org/10.1016/j.ultramic.2015.11.001", + "doc_id": "a2e8eb201167feb2", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/pii/S0304399115300644/pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Timothy L. Burnett, Timothy L. Burnett, R. Kelley, B. Winiarski, L. Contreras, Michael Daly, A. Gholinia, M. G. Burke, and P. J. Withers. Large volume serial section tomography by xe plasma fib dual beam microscopy. Ultramicroscopy, 161:119-129, Feb 2016. URL: https://doi.org/10.1016/j.ultramic.2015.11.001, doi:10.1016/j.ultramic.2015.11.001. This article has 440 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "winiarski2021correctionofartefacts", + "dockey": "5e305820c91848ff", + "citation": "B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "winiarski2021correctionofartefacts", + "bibtex": "@misc{winiarski2021correctionofartefacts,\n author = \"Winiarski, B. and Gholinia, A. and Mingard, K. and Gee, M. and Thompson, G. and Withers, P.J.\",\n title = \"Correction of artefacts associated with large area EBSD\",\n year = \"2021\",\n journal = \"Ultramicroscopy\",\n volume = \"226\",\n pages = \"113315\",\n month = \"Jul\",\n doi = \"10.1016/j.ultramic.2021.113315\",\n url = \"https://doi.org/10.1016/j.ultramic.2021.113315\",\n publisher = \"Elsevier BV\",\n issn = \"0304-3991\"\n}\n", + "authors": [ + "B. Winiarski", + "A. Gholinia", + "K. Mingard", + "M. Gee", + "G. Thompson", + "P.J. Withers" + ], + "publication_date": "2021-07-01T00:00:00Z", + "year": 2021, + "volume": "226", + "issue": null, + "issn": "0304-3991", + "pages": "113315", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2021.113315", + "title": "Correction of artefacts associated with large area EBSD", + "citation_count": 30, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2021.113315", + "doi_url": "https://doi.org/10.1016/j.ultramic.2021.113315", + "doc_id": "5e305820c91848ff", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/pii/S0304399121001017", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "B. Winiarski, A. Gholinia, K. Mingard, M. Gee, G. Thompson, and P.J. Withers. Correction of artefacts associated with large area ebsd. Jul 2021. URL: https://doi.org/10.1016/j.ultramic.2021.113315, doi:10.1016/j.ultramic.2021.113315. This article has 30 citations and is from a peer-reviewed journal." + } + ], + "call_x3Q5do7NWtrRZiM5mvwykX6y": [ + { + "embedding": null, + "docname": "folwarczny2026accurategrainboundary", + "dockey": "59d8aa55a8c3757a", + "citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "folwarczny2026accurategrainboundary", + "bibtex": "@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n", + "authors": [ + "Martin Folwarczny", + "Ao Li", + "Rushvi Shah", + "Aaron Chote", + "Alexandra C. Austin", + "Yimin Zhu", + "Gregory S. Rohrer", + "Michael A. Jackson", + "Souhardh Kotakadi", + "Katharina Marquardt" + ], + "publication_date": "2026-02-01T00:00:00Z", + "year": 2026, + "volume": "280", + "issue": null, + "issn": "0304-3991", + "pages": "114262", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "title": "Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2025.114262", + "doi_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "doc_id": "59d8aa55a8c3757a", + "file_location": null, + "license": null, + "pdf_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "stanek2020comparisonofsegmentation", + "dockey": "c163ded6c3506dd4", + "citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stanek2020comparisonofsegmentation", + "bibtex": "@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n", + "authors": [ + "J. Stanek", + "J. Kope\u010dek", + "P. Kr\u00e1l", + "I. Karafi\u00e1tov\u00e1", + "F. Seitl", + "V. Bene\u0161" + ], + "publication_date": "2020-01-01T00:00:00Z", + "year": 2020, + "volume": "58", + "issue": "05", + "issn": "0023-432X", + "pages": "301-319", + "journal": "Metallic Materials", + "publisher": "Central Library of the Slovak Academy of Sciences", + "url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=f5jvks8o3ufok0s0abmso3h0r0", + "title": "Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.4149/km_2020_5_301", + "doi_url": "https://doi.org/10.4149/km_2020_5_301", + "doc_id": "c163ded6c3506dd4", + "file_location": null, + "license": null, + "pdf_url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=buc29d44pbljhmfhjpava1l817", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "crossref", + "semantic_scholar", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations." + } + ], + "call_GpTgn8r9BIdram9mIMRTK5Lk": [], + "call_AqiInZ3vFforOEM02yK3ZwoA": [], + "call_RgOFyPWBn5U8Hvx3JRcTYJut": [], + "call_V0DFMvNN6qIFHJFPJYNQ3VAm": [], + "call_ukRTrcFmrr3QCZ78ACrh54nw": [], + "call_viEbTlkbaVWSj9viiSsVoQeq": [ + { + "embedding": null, + "docname": "shen2019convolutionalneuralnetworkbased", + "dockey": "c7144b38ae859ed1", + "citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "shen2019convolutionalneuralnetworkbased", + "bibtex": "@article{shen2019convolutionalneuralnetworkbased,\n author = \"Shen, Yu-Feng and Pokharel, Reeju and Nizolek, Thomas J. and Kumar, Anil and Lookman, Turab\",\n title = \"Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns\",\n year = \"2019\",\n journal = \"Acta Materialia\",\n month = \"May\",\n doi = \"10.1016/j.actamat.2019.03.026\",\n url = \"https://doi.org/10.1016/j.actamat.2019.03.026\",\n volume = \"170\",\n pages = \"118-131\",\n publisher = \"Elsevier BV\",\n issn = \"1359-6454\"\n}\n", + "authors": [ + "Yu-Feng Shen", + "Reeju Pokharel", + "Thomas J. Nizolek", + "Anil Kumar", + "Turab Lookman" + ], + "publication_date": "2019-05-01T00:00:00Z", + "year": 2019, + "volume": "170", + "issue": null, + "issn": "1359-6454", + "pages": "118-131", + "journal": "Acta Materialia", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "title": "Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns", + "citation_count": 67, + "bibtex_type": "article", + "source_quality": 3, + "is_retracted": null, + "doi": "10.1016/j.actamat.2019.03.026", + "doi_url": "https://doi.org/10.1016/j.actamat.2019.03.026", + "doc_id": "c7144b38ae859ed1", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1359645419301697", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Yu-Feng Shen, Reeju Pokharel, Thomas J. Nizolek, Anil Kumar, and Turab Lookman. Convolutional neural network-based method for real-time orientation indexing of measured electron backscatter diffraction patterns. Acta Materialia, 170:118-131, May 2019. URL: https://doi.org/10.1016/j.actamat.2019.03.026, doi:10.1016/j.actamat.2019.03.026. This article has 67 citations and is from a highest quality peer-reviewed journal." + } + ], + "call_sNECA4HSwXwytqITdKKFiRzx": [], + "call_gQXQyhWpcdQAtqZMTz266Aco": [], + "call_4klcSpImEsXAZeSqWkaHiygh": [], + "call_G0lw5LzH1ACLfx203yBWOksS": [], + "call_mH0dRIstCN2AOJxkl8ecCACI": [], + "call_xTeD5YsxnoGXce4YcmR9STo0": [], + "call_Ky7P0IBBulkQ2t2Ljs7dEu3k": [], + "call_ByYZSnq7nwSJDkOm9Cg48Zp8": [], + "call_sOxhMHmNJ6cGdwzj14gJbWOY": [], + "call_Dgfm5HSnCqYdsd2pIw7qCf6a": [], + "call_ERLgQRfcrgfEuXOF5b3WqhbB": [], + "call_PqUIwWt0TdvA1AJFRXJAtK32": [], + "call_Mv315OulvDSqQgHazJenPV4E": [], + "call_JvmVAiIMF8mMq3af5Jp56p6p": [], + "call_q8dUzD63NUAJ7NfyX8ac9nmd": [] + }, + "used_contexts": [ + "pqac-00000000", + "pqac-00000002", + "pqac-00000007", + "pqac-00000009", + "pqac-00000016", + "pqac-00000015", + "pqac-00000003", + "pqac-00000006", + "pqac-00000014", + "pqac-00000013", + "pqac-00000008", + "pqac-00000012", + "pqac-00000011", + "pqac-00000004", + "pqac-00000010", + "pqac-00000005", + "pqac-00000017", + "pqac-00000001" + ] + }, + "bibtex": null, + "status": "success", + "timing_info": null, + "duration": 0.0, + "stats": null, + "usage": {} + }, + "cost": 0.0, + "permission": "private", + "query_group": null + }, + "info": { + "query": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature." + } + }, + "supplemental_data": null, + "created_at": "2026-08-11T22:11:04.387655Z" + }, + "metadata": null, + "deployment_config": null, + "failure_reason": null, + "task": "Systematic prior-art survey: closed-loop, automated, or algorithm-driven optimization of EBSD (electron backscatter diffraction) data-processing and indexing parameters.\n\nSpecifically map what has been published on:\n1. Automated or optimizer-driven tuning of EBSD pattern-processing or indexing parameters (background correction, binning, neighbour averaging, Hough transform settings such as band count / rho-theta smoothing, dictionary indexing settings), using Bayesian optimization, Gaussian processes, genetic/evolutionary algorithms, grid/random search, or reinforcement learning.\n2. Feedback loops that re-index stored raw EBSD patterns and use quality metrics (confidence index CI, image quality IQ, mean angular deviation / fit, normalized cross-correlation NCC) as an objective function to select the next parameter set.\n3. Work that optimizes against GRAIN-LEVEL or microstructure-level outcomes (grain count stability vs misorientation threshold, fraction of speckle/wild grains, grain boundary integrity, fraction of pixels altered by cleanup) rather than only per-pixel pattern-level scores \u2014 i.e., multi-objective or constrained optimization of EBSD processing.\n4. Hybrid indexing strategies (Hough first, dictionary indexing on low-quality residue) and any published AUTOMATED criteria for when to escalate from Hough to dictionary/pattern-matching indexing.\n5. Software tools or frameworks claiming autonomous, self-driving, or human-out-of-the-loop EBSD post-processing or characterization workflows, including anything connecting post-processing results back to SEM acquisition-parameter recommendations (beam current, dwell time, step size).\n\nFor each area, identify the closest published work with citations, state clearly what it does and does NOT automate, and conclude with an assessment of which of the five areas remain unclaimed gaps in the literature.", + "started_at": "2026-08-11T22:00:13.577779Z", + "crow": "job-futurehouse-paperqa3", + "notification_enabled": false, + "notification_type": null, + "continued_trajectory_id": null, + "is_trajectory_owner": true, + "gcloud_operation_name": "sandbox://2a641486-1ec5-49ad-b3e8-b2836932d961/sandbox-claim-dc825b0b/job-futurehouse-paperqa3-agent-warmpool-9q9m2", + "task_summary": "Prior\u2011Art Survey Of Automated Optimization And Closed\u2011Loop EBSD Processing Methods and Gaps", + "id": "2a641486-1ec5-49ad-b3e8-b2836932d961" +} \ No newline at end of file diff --git a/outputs/edison-mtex-alternatives/_task_id.json b/outputs/edison-mtex-alternatives/_task_id.json new file mode 100644 index 0000000..fd5f34a --- /dev/null +++ b/outputs/edison-mtex-alternatives/_task_id.json @@ -0,0 +1,5 @@ +{ + "task_id": "f55d0bdd-87d8-4cf2-8d86-23309987262a", + "job": "job-futurehouse-paperqa3", + "query": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB." +} \ No newline at end of file diff --git a/outputs/edison-mtex-alternatives/answer.md b/outputs/edison-mtex-alternatives/answer.md new file mode 100644 index 0000000..fc0e42c --- /dev/null +++ b/outputs/edison-mtex-alternatives/answer.md @@ -0,0 +1,207 @@ +Question: What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB. + +# Open-source Python options for EBSD post-processing + +## Executive conclusion + +There is currently **no single Python-native package that reproduces the full integrated scope and maturity of MTEX**. The most credible MATLAB-free solution is a composed stack: + +1. **kikuchipy + PyEBSDIndex** for pattern processing and indexing; +2. **DREAM3D-NX through `simplnx`** for reproducible grain segmentation, cleanup, grain statistics, and especially 3D-capable workflows; +3. **orix** as the crystallographic orientation, symmetry, misorientation, IPF, and pole-figure layer; +4. optionally **DefDAP** for convenient 2D grain/deformation-map analysis; +5. a separately tested module for **KAM**, and especially for **GND-density inversion**. + +For the proposed automated AlSi10Mg pipeline, DREAM3D-NX/`simplnx` is the strongest grain-characterization backend, while orix is the best Python-native crystallographic foundation. kikuchipy itself is not a downstream replacement for MTEX, and pyxem is primarily a diffraction-microscopy/4D-STEM package rather than a conventional SEM-EBSD grain-analysis suite. + +| Package / ecosystem | Python-native status | Grain reconstruction / segmentation | Texture ODF / PF / IPF | KAM / GND | Maturity / reference status | Recommended role | +|---|---|---|---|---|---|---| +| MTEX | No; MATLAB-based | **Strong** for symmetry-aware 2D grain detection/reconstruction; advanced specialized workflows documented | **Strong**; benchmark for ODF, pole figures, IPF, texture components (folwarczny2026accurategrainboundary pages 5-9, savage2021anautomatedprocedure pages 1-2, savage2021anautomatedprocedure pages 3-4) | **Strong** in the MTEX ecosystem; widely treated as benchmark | Very mature, heavily cited benchmark; peer-reviewed advanced EBSD workflows documented (savage2021anautomatedprocedure pages 1-2, savage2021anautomatedprocedure pages 3-4) | Functional benchmark and validation target, but not MATLAB-free | +| kikuchipy | Yes | **Absent / not core** for downstream grain reconstruction; primarily EBSD pattern processing/indexing support | **Partial / custom** via interoperability, not a full texture-analysis suite | **Absent / not core** | Open-source Python EBSD toolbox with literature record, but not a complete post-processing replacement in the retrieved evidence | Upstream pattern handling, preprocessing, indexing, data I/O; pair with orix and/or DREAM3D-NX | +| orix | Yes | **Partial / custom**; orientation objects and misorientation logic support custom grain segmentation, but not MTEX-level turnkey reconstruction | **Strong / near-benchmark for orientation visualization**; IPF/PF native, ODF possible but typically less turnkey than MTEX | **Partial / custom**; KAM-like metrics and curvature workflows need user implementation/validation | Peer-reviewed library record available; positioned as crystallographic orientation library rather than full EBSD suite | Core Python orientation engine for custom EBSD post-processing and visualization | +| DefDAP | Yes | **Partial**; useful 2D deformation/EBSD map grain analysis, but narrower than MTEX | **Partial**; some crystallographic plotting/analysis, not a full MTEX texture environment | **Partial**; deformation-oriented metrics are a strength, GND generally custom/limited | Open-source Python project with software-release citation; maturity moderate, community smaller than MTEX/DREAM3D | Good lightweight Python choice for 2D grain/deformation analysis after indexing | +| DREAM3D-NX / simplnx | Yes (Python API around DREAM3D-NX) | **Strong**; mature pipeline-based segmentation/reconstruction, especially for 2D/3D EBSD and batch automation (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5, yi2025rapidgrainsegmentation pages 4-7, stanek2020comparisonofsegmentation pages 1-2) | **Partial**; some crystallographic statistics, but not as texture-centric or interactive as MTEX (yi2025rapidgrainsegmentation pages 4-7) | **Partial / custom**; not typically the first choice for KAM/GND-centric analysis | Mature EBSD/microstructure pipeline ecosystem with peer-reviewed reconstruction studies (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5) | Best Python-accessible option for robust automated grain segmentation/statistics in a MATLAB-free pipeline | +| pyxem | Yes | **Absent / not core for SEM-EBSD**; focused on diffraction microscopy in TEM/4D-STEM, not a direct MTEX replacement | **Partial** for orientation mapping/visualization in diffraction-microscopy contexts | **Absent / not core** | Mature open-source microscopy library, but its center of gravity is 4D-STEM/SPED rather than conventional SEM-EBSD post-processing | Use only if your workflow spans diffraction microscopy; not recommended as primary EBSD grain-texture backend | +| PyMicro | Yes | **Partial / custom**; useful microstructure data structures and crystallography, but less turnkey for EBSD grain reconstruction | **Partial / custom** | **Partial / custom** | Useful research software ecosystem, but less established as a mainstream EBSD post-processing stack in the retrieved evidence | Supporting library for custom workflows, data models, simulation coupling | +| OpenXY / research scripts / pyEBSD-style code | Yes, but heterogeneous | **Partial / custom** to **strong in isolated tasks**; highly variable and often method-specific | **Partial / custom** | **Partial / custom**, including possible GND research implementations | Lowest standardization; often lacks broad peer-reviewed software validation or stable APIs | Fill specific gaps such as KAM/GND or custom segmentation, but validate carefully against MTEX/DREAM3D outputs | +| Practical no-MATLAB stack for AM AlSi10Mg | Yes | **Strong overall** if composed as: kikuchipy/PyEBSDIndex + DREAM3D-NX for segmentation + orix/DefDAP for orientation/statistics | **Partial-to-strong overall**; strongest with orix plus custom ODF tooling | **Partial / custom overall**; KAM feasible, GND should be a separately validated module | Most realistic MATLAB-free path based on current open-source ecosystem and evidence balance (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5) | Recommended integrated pipeline rather than searching for a single Python package that fully replaces MTEX | + + +*Table: This table compares major open-source Python-native or Python-accessible EBSD post-processing options against MTEX for grain reconstruction, texture analysis, and misorientation-derived metrics. It is designed to help choose a MATLAB-free downstream pipeline for automated EBSD workflows.* + +## The MTEX benchmark + +MTEX combines, in one environment, symmetry-aware EBSD import and manipulation, grain reconstruction, grain and boundary objects, texture estimation and plotting, misorientation statistics, KAM/GROD-type measures, and curvature/dislocation-density workflows. Its texture functionality includes ODF estimation, pole and inverse-pole figures, texture components, fibres, and quantitative texture measures. Published applications use MTEX for IPF maps, ODF-derived pole figures, grain-size and grain-shape statistics, and grain-boundary analysis. (folwarczny2026accurategrainboundary pages 5-9) + +MTEX also supports sophisticated extensions beyond simple threshold segmentation. A peer-reviewed workflow built on MTEX and MATLAB graph tools reconstructed deformation-twin hierarchies, grouped deformed fragments into parent families, treated missing relationships, and identified third-generation twins in heavily twinned α-Ti. (savage2021anautomatedprocedure pages 1-2) This illustrates the principal gap in Python: individual Python libraries provide many underlying operations, but MTEX supplies a more coherent grain/boundary/texture abstraction and a larger collection of specialized reconstruction workflows. + +MTEX is not immune to methodological choices. Strong textures and low-angle boundaries can cause under-segmentation or chains of crystallographically similar regions; one MTEX study deliberately biased reconstruction toward over-segmentation and used 1–2° tolerances for its particular material. (savage2021anautomatedprocedure pages 3-4) More generally, grain counts and geometry depend strongly on the selected threshold, connectivity, cleanup, and minimum-grain-size rules. A conventional 5° threshold is common but is not universally valid. (stanek2020comparisonofsegmentation pages 1-2) + +## Package assessment + +### 1. kikuchipy + +**Role.** kikuchipy is the natural upstream package in the proposed workflow. It is designed around EBSD patterns: vendor-file access, lazy handling of large pattern arrays, detector and projection geometry, background correction, pattern processing, simulation-related operations, and indexing interoperability. Its data model uses the HyperSpy ecosystem and orix crystallographic objects. + +**Grains and texture.** It should not be selected as the principal grain-reconstruction engine. Once indexing has produced orientation, phase, quality, and coordinate arrays, downstream segmentation and grain statistics generally require another package or custom code. Likewise, kikuchipy is not a comprehensive ODF/texture-analysis system and does not offer an MTEX-equivalent grain/boundary object model. + +**KAM/GND.** Neither is a core turnkey kikuchipy capability. KAM can be calculated from its output arrays with orix/NumPy-style operations, but masking, symmetry reduction, neighborhood definition, and exclusion of high-angle boundaries remain the pipeline developer’s responsibility. GND inversion is substantially further outside its main scope. + +**Maturity.** kikuchipy is a well-developed, domain-specific open-source project with a software paper titled *kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns*. Its maturity is strongest for pattern-space operations, not for post-indexing microstructure analysis. + +### 2. orix + +**Role.** orix is the most important Python-native crystallographic foundation. It supplies rotations and orientations, crystal and specimen symmetry, fundamental-zone reduction, symmetry-aware misorientation, orientation colours, inverse-pole-figure colour keys, pole-figure-style projections, and orientation clustering. Its peer-reviewed methodological reference is Johnstone et al., *Density-based clustering of crystal (mis)orientations and the orix Python library*, *Journal of Applied Crystallography* 53 (2020), 1293–1298, DOI 10.1107/S1600576720011103. + +**Grain reconstruction.** orix contains the primitives needed to compare neighboring orientations and cluster orientation data, but it is not yet an MTEX-like turnkey EBSD topology package. A production grain segmenter still needs scan-grid connectivity, phase masks, confidence masks, flood-fill or graph labeling, small-region handling, and grain-property aggregation. Those steps can be implemented with NumPy, SciPy, scikit-image, xarray, and networkx, but then the laboratory owns validation and maintenance. + +**Texture.** orix is strong for IPF colouring, stereographic projections, symmetry handling, and discrete orientation distributions. It is suitable for PF/IPF visualization and custom orientation-density estimation. Nevertheless, MTEX remains substantially richer and more established for quantitative ODF estimation, kernel-selection conventions, harmonic representations, fibres/components, uncertainty-aware texture workflows, and publication-ready texture analysis. An ODF reconstructed from discrete EBSD orientations is also parameter-dependent; published work notes sensitivity to ODF computation settings in MTEX itself, so Python/MTEX agreement requires matching kernels, bandwidths, symmetry, weights, and normalization. A modern EBSD texture study used pole figures to characterize weak texture and illustrates the need to retain a statistically representative orientation population. (wanni2024machinelearningenhanced pages 1-2) + +**KAM/GND.** orix is a good basis for KAM because it provides symmetry-aware misorientation operations. It does not, by itself, turn KAM into a one-call validated EBSD workflow. GND requires spatial orientation gradients, reference-frame consistency, Burgers vectors/slip-system assumptions, Nye-tensor construction, and an underdetermined inversion; this remains custom research code rather than an MTEX-equivalent orix feature. + +### 3. DefDAP + +**Role.** DefDAP—Deformation Data Analysis in Python—is a Python package aimed at combining EBSD with deformation measurements such as high-resolution digital image correlation. Its citable software release is Atkinson et al., *DefDAP: Deformation Data Analysis in Python* (2020), DOI 10.5281/zenodo.3688097. + +**Grains.** DefDAP is useful for 2D EBSD maps, grain finding and grain-level analysis, crystallographic slip/twinning interpretation, boundary overlays, and linking deformation fields to grains. For a conventional single-phase 2D Al EBSD map it may be the quickest lightweight route from an orientation map to grains and grain-wise quantities. + +**Texture and misorientation metrics.** Its emphasis is local deformation and grain-scale mechanics, not comprehensive quantitative texture. It is less complete than MTEX for ODFs, texture components, and generalized boundary/reconstruction operations. KAM-like local misorientation calculations can be added or may be available through map-level operations depending on release, but GND-density inversion should not be assumed to be a complete, standardized capability without checking the exact version and validating the implementation. + +**Maturity.** DefDAP has been used in peer-reviewed deformation studies, but its community, documentation breadth, and EBSD feature coverage are smaller than MTEX or DREAM.3D. It is best regarded as a focused 2D deformation-analysis package rather than a universal EBSD backend. + +### 4. DREAM.3D and DREAM3D-NX/`simplnx` + +**Role and architecture.** Legacy DREAM.3D is an established C++/Qt microstructure-analysis platform. DREAM3D-NX is its current-generation successor; `simplnx` provides Python bindings and pipeline construction. Thus, it is **Python-accessible and scriptable**, although not pure Python internally. This distinction is normally beneficial for large EBSD maps because segmentation and feature-statistics filters execute in compiled code. + +**Grain reconstruction.** This is the strongest open-source alternative for automated grain segmentation. DREAM.3D uses pipeline filters for data import, masking, cleanup, symmetry-aware misorientation segmentation, feature identification, neighbor lists, grain morphology, crystallographic statistics, boundary meshes, and, where applicable, 3D reconstruction. Published work describes DREAM.3D as a unified data structure aggregating community-developed algorithms, with reconstruction pipelines commonly containing more than a dozen parameterized steps. (polonsky2019threedimensionalanalysisand pages 1-2) + +Its `Segment Features (Misorientation)` workflow has been used to compare 2D and 3D segmentation across several misorientation thresholds. (stanek2020comparisonofsegmentation pages 2-7) Published AM/TriBeam workflows reconstructed serial EBSD slices, segmented grains using explicit misorientation tolerances, removed features below specified voxel counts, filled gaps, and automated parameter sweeps. (polonsky2019threedimensionalanalysisand pages 2-5) Cloud integration further demonstrated parallel parameter studies, provenance tracking, and grain-size comparisons between 2D and 3D AM datasets. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14) + +**Texture.** DREAM.3D calculates crystallographic feature statistics and supports orientation-distribution data used in synthetic microstructure generation, but it is not as strong as MTEX for interactive quantitative texture analysis. PF/IPF maps and basic orientation statistics are feasible; sophisticated ODF reconstruction and texture-component analysis are better handled in orix plus custom numerical code—or validated against MTEX. Published workflows frequently use DREAM.3D for microstructure generation/reconstruction and MTEX for detailed ODF/PF analysis, reflecting this division of strengths. (folwarczny2026accurategrainboundary pages 5-9) + +**KAM/GND.** DREAM3D-NX should not be selected primarily for KAM or GND. Depending on installed plugins and version, local orientation filters may cover some neighborhood measures, but these capabilities are less central and less transparent than its feature segmentation/statistics pipeline. A NumPy/orix KAM stage is easier to audit. GND remains a specialist calculation. + +**Cautions.** Segmentation is not parameter-free. The same threshold can produce different 2D and 3D connectivity, especially for nonconvex grains or paths connected outside the selected plane. (stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 2-7) Minimum feature size and cleanup can also shift the measured grain-size distribution. Consequently, pipeline JSON, threshold, neighborhood convention, phase mask, and deleted/fill operations should be retained as provenance. + +### 5. pyxem + +pyxem is an open-source Python ecosystem for multidimensional diffraction microscopy, built around HyperSpy and commonly used for scanning/precession electron diffraction and 4D-STEM. It is valuable for virtual diffraction imaging, diffraction-vector analysis, orientation mapping, and related TEM workflows. + +It is **not a direct downstream package for conventional SEM-EBSD orientation maps**. Grain segmentation from vendor-style EBSD grids, quantitative EBSD texture analysis, KAM, and GND are not its central abstractions. Its shared ecosystem with kikuchipy/orix can be useful if the project later combines EBSD with scanning electron diffraction, but pyxem should not be the primary AlSi10Mg grain-analysis backend. + +### 6. PyMicro + +PyMicro provides Python classes and algorithms for crystallography, microstructure representation, grains, orientations, meshes, and links to simulation and tomography. It can import or represent EBSD-like microstructures and is useful for custom data integration, synthetic microstructures, and computational-mechanics coupling. + +For this task, its limitations are that conventional EBSD segmentation, comprehensive ODF analysis, KAM, and GND are not offered with the same turnkey breadth or community validation as MTEX. It is better viewed as an extensible microstructure framework than as a drop-in EBSD analysis application. + +### 7. OpenXY and other research code + +OpenXY is associated mainly with high-angular-resolution EBSD cross-correlation and strain/rotation analysis. It addresses a different problem from ordinary Hough-indexed orientation-map characterization and is not a complete grain/texture package. Some Python ports and research repositories calculate lattice curvature or GND density, but their conventions, slip-system bases, regularization, and maintenance vary. + +Other small packages carrying names such as `pyebsd`, custom scikit-image segmenters, or laboratory scripts can calculate IPF colours, KAM, or threshold-connected grains. They can be useful building blocks, but package names are sometimes ambiguous, releases may be unmaintained, and peer-reviewed software references are often absent. They should not be treated as equivalent to MTEX merely because they produce an IPF or KAM image. + +## Recommended pipeline for AM AlSi10Mg + +### Data contract after indexing + +Export or retain the following arrays and metadata: + +- quaternion or rotation matrix per scan point—preferably not only Euler angles; +- phase ID, indexed/unindexed mask, confidence or match score, and pattern quality; +- physical `x`, `y` coordinates and step size; +- scan shape and rectangular/hexagonal topology; +- crystal symmetry and lattice parameters for fcc Al; +- detector/sample/reference-frame conventions; +- indexing dictionary or phase-library version. + +AlSi10Mg is commonly dominated in EBSD by fcc α-Al, while the fine Si network may be incompletely indexed at ordinary step sizes. Do not let unindexed Si-rich pixels automatically split Al grains: establish an explicit phase/mask and cleanup policy and report it. + +### Grain stage + +Use **DREAM3D-NX/`simplnx`** for the production pipeline: + +1. construct the image geometry and load quaternion, phase, mask, and quality arrays; +2. remove only demonstrably unreliable points; +3. segment same-phase connected pixels with a documented disorientation threshold; +4. calculate feature centroids, equivalent diameters, areas, aspect ratios, mean orientations, neighbors, and boundary misorientations; +5. retain both the raw feature labels and any post-cleanup labels; +6. sweep plausible thresholds rather than fixing one unexamined value. + +For AM AlSi10Mg, evaluate at least a low-angle-sensitive threshold and a conventional grain threshold—for example, a controlled sweep around 2–5°—because cellular/subgrain orientation gradients may be physically meaningful. This is not a recommendation to choose a universal value; published work demonstrates that changing tolerance can radically alter segmentation and that 2D and 3D results are not automatically equivalent. (polonsky2019threedimensionalanalysisand pages 2-5, stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 2-7) + +If only modest 2D maps are involved and direct Python object manipulation is more important than DREAM3D-NX pipelines, **DefDAP** is a reasonable alternative. A fully custom orix/scikit-image segmenter is defensible only if regression-tested against DREAM3D-NX or MTEX. + +### Texture stage + +Use **orix** for orientation conversion, symmetry reduction, IPF maps, pole figures, and discrete orientation statistics. For ODF estimation: + +- weight points or grains deliberately; pixel-weighted and one-orientation-per-grain ODFs answer different questions; +- state the kernel and half-width/bandwidth; +- enforce cubic crystal symmetry and the intended specimen symmetry; +- normalize PF/ODF intensities consistently, usually in multiples of random distribution; +- validate against a known texture or a small MTEX reference dataset. + +This distinction matters in AM material: pixel weighting emphasizes large columnar grains, whereas grain weighting describes the population of grains. A recent study highlights the importance of preserving the full orientation distribution when reducing EBSD orientation datasets and uses PF maps to assess texture strength. (wanni2024machinelearningenhanced pages 1-2) + +### KAM + +KAM is straightforward to implement but easy to define inconsistently. For each indexed point, compute the mean symmetry-reduced disorientation to valid neighbors, while specifying: + +- first-, second-, or higher-order neighborhood; +- whether diagonals are included; +- maximum included disorientation, often chosen to exclude grain boundaries; +- treatment of phase boundaries, map edges, and unindexed pixels; +- degrees versus radians; +- whether the map has been denoised before calculation. + +Use orix for disorientation and NumPy/SciPy for neighborhood operations. Store the number of valid neighbors along with KAM; otherwise values near pores, Si particles, and map boundaries are misleading. KAM should not be interpreted directly as dislocation density. + +### GND density + +GND is the least turnkey part of a MATLAB-free pipeline. A defensible calculation needs orientation gradients in physical units, lattice curvature, an explicit Nye-tensor convention, Burgers-vector/slip-system definitions, and a solution to an underdetermined inversion—commonly L1 or L2 minimization. Report whether the result is total scalar GND density, a lower bound, or slip-system-resolved density. + +For conventional EBSD, noise and angular resolution strongly affect spatial derivatives. The result also scales with step size and preprocessing. Therefore: + +1. preserve orientations at full precision; +2. estimate angular noise on an undeformed/reference region; +3. document derivative stencil and smoothing; +4. perform step-size or downsampling sensitivity tests; +5. benchmark synthetic rotation-gradient fields with known curvature; +6. compare a representative map against an established MTEX or HR-EBSD implementation. + +No named Python package in this comparison offers a broadly validated, turnkey GND workflow matching MTEX’s integration. Treat GND as an independently versioned scientific module, not as a plotting option attached to the grain segmenter. + +## Practical ranking + +- **Best production grain segmentation and statistics:** DREAM3D-NX/`simplnx`. +- **Best Python crystallographic foundation and PF/IPF layer:** orix. +- **Best upstream pattern/indexing integration:** kikuchipy + PyEBSDIndex. +- **Best lightweight 2D grain/deformation package:** DefDAP. +- **Best for TEM diffraction-microscopy integration, not SEM-EBSD post-processing:** pyxem. +- **Best custom microstructure/simulation framework:** PyMicro. +- **No fully satisfactory turnkey Python winner for GND:** implement and validate separately. + +Accordingly, the recommended deployable stack is **kikuchipy/PyEBSDIndex → DREAM3D-NX/`simplnx` → orix-based texture and KAM → validated custom GND module**. This removes MATLAB from routine production while retaining MTEX only as an optional one-time validation reference. DREAM.3D’s documented use on complex additively manufactured 3D EBSD datasets, including automated parameter studies and explicit misorientation segmentation, makes this architecture substantially lower-risk than building the entire downstream stage directly from NumPy/orix primitives. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5) + +References + +1. (folwarczny2026accurategrainboundary pages 5-9): Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal. + +2. (savage2021anautomatedprocedure pages 1-2): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal. + +3. (savage2021anautomatedprocedure pages 3-4): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal. + +4. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +5. (polonsky2019threedimensionalanalysisand pages 13-14): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +6. (polonsky2019threedimensionalanalysisand pages 2-5): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +7. (yi2025rapidgrainsegmentation pages 4-7): Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal. + +8. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. + +9. (wanni2024machinelearningenhanced pages 1-2): J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4. This article has 14 citations and is from a peer-reviewed journal. + +10. (stanek2020comparisonofsegmentation pages 2-7): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. \ No newline at end of file diff --git a/outputs/edison-mtex-alternatives/references.txt b/outputs/edison-mtex-alternatives/references.txt new file mode 100644 index 0000000..2e2d188 --- /dev/null +++ b/outputs/edison-mtex-alternatives/references.txt @@ -0,0 +1,19 @@ +1. (folwarczny2026accurategrainboundary pages 5-9): Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal. + +2. (savage2021anautomatedprocedure pages 1-2): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal. + +3. (savage2021anautomatedprocedure pages 3-4): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal. + +4. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +5. (polonsky2019threedimensionalanalysisand pages 13-14): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +6. (polonsky2019threedimensionalanalysisand pages 2-5): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal. + +7. (yi2025rapidgrainsegmentation pages 4-7): Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal. + +8. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. + +9. (wanni2024machinelearningenhanced pages 1-2): J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4. This article has 14 citations and is from a peer-reviewed journal. + +10. (stanek2020comparisonofsegmentation pages 2-7): J. Stanek, J. Kopeček, P. Král, I. Karafiátová, F. Seitl, and V. Beneš. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\_2020\_5\_301, doi:10.4149/km\_2020\_5\_301. This article has 4 citations. \ No newline at end of file diff --git a/outputs/edison-mtex-alternatives/task_full.json b/outputs/edison-mtex-alternatives/task_full.json new file mode 100644 index 0000000..5c52e2e --- /dev/null +++ b/outputs/edison-mtex-alternatives/task_full.json @@ -0,0 +1,6278 @@ +{ + "status": "success", + "query": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "user": null, + "created_at": "2026-07-29T03:32:08.878330Z", + "job_name": "job-futurehouse-paperqa3", + "share_status": "private", + "permitted_accessors": { + "users": [], + "organizations": [] + }, + "build_owner": "FutureHouse", + "environment_name": null, + "agent_name": null, + "task_id": "f55d0bdd-87d8-4cf2-8d86-23309987262a", + "project_id": null, + "agent_state": [ + { + "id": "40806201-7df8-4191-a1df-3dd1fe314e9e", + "trajectory_id": "f55d0bdd-87d8-4cf2-8d86-23309987262a", + "trajectory_timestep": 14, + "step": "after_transition", + "state": { + "transition": { + "timestep": 14, + "agent_state": { + "tools": [ + { + "type": "function", + "info": { + "name": "create_plan", + "description": "Create a plan for to construct a thorough and accurate answer.\n\nDecompose the input task into a set of sub-objectives that need to\nbe met in order to form the answer. Subsequently, when working on solving\nthe query, you can focus on these objectives one by one. Note that the\n`update_plan` tool allows you to change the objectives if needed.\n\nIn addition to the objectives you deem necessary, add one final objective\nto call the `answer` tool with your final answer, so that you do not\nforget to submit your answer.", + "parameters": { + "type": "object", + "properties": { + "plan_json": { + "description": "A JSON string representing the plan. It should be a dictionary\nwith top-level key \"objectives\", whose value is a list of objectives.\nEach objective should be a dictionary with the following keys:\n- \"id\": (int) an integer id for the plan objective\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Plan Json", + "type": "string" + } + }, + "required": [ + "plan_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "patent_search", + "description": "Search for patents to increase the patent count.\n\nRepeat previous calls with the same query to continue a search.\nOnly repeat a maximum of three times.\nThis tool can be called concurrently.\nThis tool introduces novel patents, so invoke it when just beginning\nor when unsatisfied with the current evidence.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A search query for patents (using Google Patents). Never include temporal\nfilters here (e.g. specific years or dates); instead use the `after` and `before`\narguments. Never include author filters here; instead use the `inventor` argument.\nSupports the following syntax, but they can be brittle and are not preferred; try\nsimple queries where possible:\n- Boolean operators: AND (default), OR, and - (NOT)\n- Exact phrases: \"quoted text\"\n- Proximity: NEAR/x, ADJ/x (ordered), WITH (20 words), SAME (200 words)\n- Field search: TI=(title), AB=(abstract), CL=(claims)\n- CPC codes: cpc:A01B or CPC=B60R22/low (includes children)\n- Wildcards: ? (0-1 char), * (0+ chars), # (exactly 1 char)\n- Chemistry: exact molecule names, SSS=molecule (substructure),\n ~molecule (similar), SMILES strings, SMARTS=pattern, InChI keys,\n CL=~molecule (chemistry in claims)", + "title": "Query", + "type": "string" + }, + "after": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Minimum patent date. Format: YYYYMMDD.\nExample: '20150101' for patents after Jan 1, 2015.", + "title": "After" + }, + "before": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Maximum patent date. Format: YYYYMMDD.\nExample: '20200101' for patents before Jan 1, 2020.", + "title": "Before" + }, + "date_type": { + "default": "publication", + "description": "Type of date that after and before refer to. Must be either 'publication'\nor 'filing'. Only necessary to set if after and/or before are set.", + "title": "Date Type", + "type": "string" + }, + "inventor": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Filter by inventor name(s). Comma-separated for multiple.", + "title": "Inventor" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "gather_evidence", + "description": "Given a specific question, gather evidence from full-text of documents that have been retrieved by\nprevious tool calls.\n\nHere is how the tool works: the provided question is posed to all texts retrieved thus far\nby other tools. An LLM sees each text independently and is prompted to find information relevant\nto the question.\n\nThat is, we sample evidence `e \\sim LLM(prompt, text, q)` for each text. We only return pieces of\nevidence `e` that are sufficiently relevant to the question; if the text is irrelevant, then `e`\nis not included in the final response of this tool. Note that this LLM does not see the conversation\nhistory thus far, only the question and each text. Note that it only sees text, no images or other\nmedia from the contents (though tables and formulae may be parseable in some cases).\n\nThink of this tool as an efficient way to deeply scan all retrieved documents for evidence. It\nis complementary to the `read_text` tool, which can be used to directly inspect a subset of the\nretrieved documents.\n\nA valuable time to invoke this tool is after another tool finds potentially relevant documents.\nFeel free to invoke this tool in parallel with other tools, but do not call this tool in parallel with itself.\nOnly invoke this tool when the document count is above zero, or this tool will be useless.\n\nIf you call this tool 2+ times and repeatedly get 'No relevant evidence found',\nconsider whether the answer might be in visual content (figures/tables) rather than text,\nand use appropriate tools if available to access that content.\n\nOnce you receive evidence from this tool:\n\n1. Read Carefully: Don't skim. The evidence may contain multiple similar terms\n or values\u2014make sure you identify the RIGHT one.\n\n2. For Comparative Questions: If the question asks \"which X has greatest Y\",\n check that the evidence includes information about ALL relevant X options, not\n just one or two. If you only see partial information, gather more evidence.\n\n3. Check for Noise: Verify that numbers or terms aren't just metadata\n (page numbers, reference IDs, dates). Look for the value in meaningful\n scientific context.\n\n4. Context Matters: A term appearing 71 times doesn't mean it's always the answer.\n Verify which occurrence actually answers the question asked.\n\nIf the evidence doesn't clearly answer the question, gather more evidence before\nconcluding. Don't guess based on partial information.", + "parameters": { + "type": "object", + "properties": { + "question": { + "description": "Specific question to gather evidence for.", + "title": "Question", + "type": "string" + }, + "text_name_focus_list": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "if there are known text names that are relevant to the question,\nyou can specify them here. This will ensure that the tool will look at these text names\nfirst, and then look at the rest of the evidence. If left as `None`, an embedding\nsearch will be performed on underlying content to find most relevant texts.", + "title": "Text Name Focus List" + } + }, + "required": [ + "question" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "update_plan", + "description": "Update the plan for a specific or new objective using a JSON string to create or\nreplace the existing objective.\n\nThe updated_json should be a JSON string representing a single plan objective object.\n\nYou may use this tool to update an objective, create a new one, or change\nstrategy if the current plan is not working. Only change strategy significantly\nas a last resort, after trying very hard to make the current plan work.\n\nOnly mark an objective as completed if you are absolutely sure that it has been fully\nmet. If it is only partially met, you may add additional objectives to the plan for\nwhat is missing.", + "parameters": { + "type": "object", + "properties": { + "plan_objective_id": { + "description": "The id of the objective to update.", + "title": "Plan Objective Id", + "type": "integer" + }, + "updated_plan_objective_json": { + "description": "A JSON string representing the updated objective.\nIt should be a dictionary with the with the following keys:\n- \"id\": (int) an integer id matching plan_objective_id, if not found, a new objective is created.\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Updated Plan Objective Json", + "type": "string" + } + }, + "required": [ + "plan_objective_id", + "updated_plan_objective_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "paper_search", + "description": "Search for papers in Google Scholar for you to look at.\n\nCan be called repeatedly with identical parameters to retrieve more results (max 2 repeats).\nA short summary of the retrieved papers will be returned, but to examine the full content of each paper,\nyou must use the gather_evidence tool.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A Google Scholar search query, which can be a specific phrase, complete sentence,\nor general keywords, e.g. 'machine learning for immunology'. Also can be\ngiven search operators like 'author:'.", + "title": "Query", + "type": "string" + }, + "min_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for minimum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Min Year" + }, + "max_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for maximum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Max Year" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "clinical_trials_search", + "description": "Search for clinical trials, with support for repeated calls and concurrent execution.\n\nWill add new clinical trials to the state, and return metadata about the number of trials found.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "The search query string. Supports complex boolean expressions, field-specific\nsearches, and query modifiers through operators. All configuration is done through\noperators in the query string.\nQuery Syntax:\n Basic Search:\n Simple text automatically uses default EXPANSION[Relaxation] and COVERAGE[Contains]\n >>> \"heart attack\"\n\n Modified Search:\n Use operators to modify search behavior:\n >>> 'EXPANSION[None]COVERAGE[FullMatch]\"exact phrase\"'\n >>> 'EXPANSION[Concept]heart attack'\n\n Field Search:\n Specify fields using AREA operator:\n >>> 'AREA[InterventionName]aspirin'\n >>> 'AREA[Phase]PHASE3'\n\n Location Search:\n Use SEARCH operator for compound location queries:\n >>> 'cancer AND SEARCH[Location](AREA[LocationCity]Boston AND AREA[LocationState]Massachusetts)'\n\n Complex Boolean:\n Combine terms with AND, OR, NOT and parentheses:\n >>> '(cancer OR tumor) AND NOT (EXPANSION[None]pediatric OR AREA[StdAge]CHILD)'\n\n Date Ranges:\n Use RANGE to specify date ranges with formats like \"yyyy-MM\" or \"yyyy-MM-dd\".\n Note that MIN and MAX can be used for open-ended ranges:\n >>> AREA[ResultsFirstPostDate]RANGE[2015-01-01, MAX]\n\nOperators:\n EXPANSION[type]: Controls term expansion\n - None: Exact match only, case and accent sensitive\n - Term: Includes lexical variants (plurals, spellings)\n - Concept: Includes UMLS synonyms\n - Relaxation: Relaxes adjacency requirements (default)\n - Lossy: Allows missing partial terms\n\n COVERAGE[type]: Controls text matching\n - FullMatch: Must match entire field\n - StartsWith: Must match beginning of field\n - EndsWith: Must match end of field\n - Contains: Must match part of field (default)\n\n AREA[field]: Specifies field to search\n - See Field Reference for available fields\n\n SEARCH[type]: Groups field searches\n - Location: Groups location-related fields\n - Study: Groups study-related fields\n\nUsage Notes:\n - All search expressions are implicitly OR expressions\n - Operator precedence (highest to lowest): terms/source operators, NOT/context operators, AND, OR\n - Use quotes for exact phrase matching: \"heart attack\"\n - Use parentheses for grouping: (heart OR cardiac) AND attack\n - Use backslash to escape operators: \\AND\n - Default expansion is EXPANSION[Relaxation]\n - Default coverage is COVERAGE[Contains]\n\nField Reference:\n High Priority Fields (weight >= 0.8):\n - NCTId (1.0): Trial identifier\n - Acronym (1.0): Study acronym\n - BriefTitle (0.89): Short title\n - OfficialTitle (0.85): Full official title\n - Condition (0.81): Medical condition\n - InterventionName (0.8): Primary intervention name\n - OverallStatus: Trial status\n\n Medium Priority Fields (0.5-0.79):\n - InterventionOtherName (0.75): Alternative intervention names\n - Phase (0.65): Trial phase\n - StdAge (0.65): Standard age groups\n - Keyword (0.6): Study keywords\n - BriefSummary (0.6): Short description\n - SecondaryOutcomeMeasure (0.5): Secondary outcomes\n\n Low Priority Fields (< 0.5):\n - DesignPrimaryPurpose (0.3): Primary purpose of study\n - StudyType (0.3)\n - Various descriptive, location, and administrative fields\n\nSupported Enums:\n Phase:\n - EARLY_PHASE1: Early Phase 1\n - PHASE1: Phase 1\n - PHASE2: Phase 2\n - PHASE3: Phase 3\n - PHASE4: Phase 4\n - NA: Not Applicable\n\n StandardAge:\n - CHILD: Child\n - ADULT: Adult\n - OLDER_ADULT: Older Adult\n\n Status:\n - RECRUITING: Currently recruiting participants\n - ACTIVE_NOT_RECRUITING: Active but not recruiting\n - COMPLETED: Study completed\n - ENROLLING_BY_INVITATION: Enrolling by invitation only\n - NOT_YET_RECRUITING: Not yet recruiting\n - SUSPENDED: Study suspended\n - TERMINATED: Study terminated\n - WITHDRAWN: Study withdrawn\n - AVAILABLE: Available\n - NO_LONGER_AVAILABLE: No longer available\n - TEMPORARILY_NOT_AVAILABLE: Temporarily not available\n - APPROVED_FOR_MARKETING: Approved for marketing\n - WITHHELD: Withheld\n - UNKNOWN: Unknown status\n\n StudyType:\n - INTERVENTIONAL: Interventional studies\n - OBSERVATIONAL: Observational studies\n - EXPANDED_ACCESS: Expanded access studies\n\n PrimaryPurpose:\n - TREATMENT: Treatment\n - PREVENTION: Prevention\n - DIAGNOSTIC: Diagnostic\n - ECT: Educational/Counseling/Training\n - SUPPORTIVE_CARE: Supportive Care\n - SCREENING: Screening\n - HEALTH_SERVICES_RESEARCH: Health Services Research\n - BASIC_SCIENCE: Basic Science\n - DEVICE_FEASIBILITY: Device Feasibility\n - OTHER: Other\n\n InterventionType:\n - BEHAVIORAL: Behavioral interventions\n - BIOLOGICAL: Biological interventions\n - COMBINATION_PRODUCT: Combination product interventions\n - DEVICE: Device interventions\n - DIAGNOSTIC_TEST: Diagnostic test interventions\n - DIETARY_SUPPLEMENT: Dietary supplement interventions\n - DRUG: Drug interventions\n - GENETIC: Genetic interventions\n - PROCEDURE: Procedure interventions\n - RADIATION: Radiation interventions\n - OTHER: Other interventions\n\n DesignAllocation:\n - RANDOMIZED: Randomized allocation\n - NON_RANDOMIZED: Non-randomized allocation\n - NA: Not applicable\n\n InterventionalAssignment:\n - SINGLE_GROUP: Single group assignment\n - PARALLEL: Parallel assignment\n - CROSSOVER: Crossover assignment\n - FACTORIAL: Factorial assignment\n - SEQUENTIAL: Sequential assignment\n\n ObservationalModel:\n - COHORT: Cohort\n - CASE_CONTROL: Case-Control\n - CASE_ONLY: Case-Only\n - CASE_CROSSOVER: Case-Crossover\n - ECOLOGIC_OR_COMMUNITY: Ecologic or Community\n - FAMILY_BASED: Family-Based\n - DEFINED_POPULATION: Defined Population\n - NATURAL_HISTORY: Natural History\n - OTHER: Other\n\n DesignMasking:\n - NONE: None (Open Label)\n - SINGLE: Single\n - DOUBLE: Double\n - TRIPLE: Triple\n - QUADRUPLE: Quadruple\n\n WhoMasked:\n - PARTICIPANT: Participant\n - CARE_PROVIDER: Care Provider\n - INVESTIGATOR: Investigator\n - OUTCOMES_ASSESSOR: Outcomes Assessor", + "title": "Query", + "type": "string" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "view_images", + "description": "Retrieve images from a document.\n\nUse this tool when you need to see figures, tables, or other visual content.\nThe tool works by asking a weaker LLM agent to scan through the document, looking for\nvisual content matching the query you provide. When the agent finds matches, it will\ncrop the page to the relevant region. It can be tasked to return multiple images,\nif desired. Note that the weaker LLM may be worse at *understanding* images than\nyou, so double-check its conclusions about the returned images. It is also stateless,\ni.e. it will not remember what you have asked it to find in the past.\n\nUse this tool carefully. It is difficult to retrieve images, and the images\nthemselves will fill up your context window quickly. Here are some guidelines:\n1. ALWAYS use this tool if the question explicitly mentions 'Table X' or 'Figure Y'.\n These questions require visual content and cannot be answered with text alone.\n2. Use this tool as a fallback when gather_evidence fails repeatedly (2+ times\n returning 'No relevant evidence'). This suggests the answer may be in visual content.\n3. When gather_evidence returns text mentioning 'see Figure X' or 'shown in Table Y',\n you should call this tool to examine that specific visual reference.\n4. Only call this tool after directly inspecting relevant text with read_text\n to confirm the document likely contains the visual you need.\n5. Don't use this tool in a broad-spectrum way (don't call on many documents\n just to see what might be relevant).\n6. Never call this tool in parallel with itself.\n7. Users like visual evidence, so call this tool at least once before answering so you\n can provide some image citations in the answer. Add this to your plan so you do not forget.\n For images to be presented to the user, they must be cited in the final answer after retrieval\n by this tool. You may ignore this instruction if the request is for the question is asking\n for a specific piece of information and is not an open-ended research question.\n8. NEVER cite full-page images in your final answer, unless they are full-page panels figures or\n tables. Cite relevant cropped regions if available, otherwise consider calling this tool again\n to find a smaller region that answers the query.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "Query to search for visual content in the document. If you know the specific\nfigure/table/formula you are looking for, you can specify it here (e.g. \"find me\nFigure 3\" or \"extract the table on page 10\"). You can also request based on content\n(e.g. \"find me all volcano plots of gene X\" or \"I need the formula for the growth rate\").\nYou can even include explicit instructions for the LLM agent to follow (if helpful, not\nnecessarily required). The more specific and clear your query is here, the higher the\nchance of success.", + "title": "Query", + "type": "string" + }, + "text_name": { + "description": "Name of a text chunk (e.g., 'smith2023example pages 1-5').", + "title": "Text Name", + "type": "string" + } + }, + "required": [ + "query", + "text_name" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "read_text", + "description": "Given a list of text names, read their contents.\n\nThis tool is valuable to read the full contents of specific parts of documents,\ncomplementing the `gather_evidence` tool's indirect approach.\nThis can be an especially useful tool if `gather_evidence` does not find evidence,\nbut you think it is very likely that specific documents are still relevant to the question.", + "parameters": { + "type": "object", + "properties": { + "text_name_list": { + "description": "A list of up to 10 text names to read.\nThese text names can be any valid text names retrieved by other tools.", + "items": { + "type": "string" + }, + "title": "Text Name List", + "type": "array" + } + }, + "required": [ + "text_name_list" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "create_artifact", + "description": "Create an artifact that can be used in the final answer, based on information\nyou have collected so far.\n\nHow it works: the full message history thus far is passed to another LLM,\nalong with the question and your description. The LLM is tasked to create\nan artifact that matches your specifications, including a title, caption,\nand markdown content. You drive the process by specifying the type and\ndescribing what you want.\n\nOnly use this tool *after* you have gathered sufficient evidence or\ninformation to create a new artifact. Do not call this tool in parallel\nwith tools that are searching for information or evidence that would be\nnecessary to create this artifact.\n\nIt is very helpful after calls to several search tools,\nafter calls to the `gather_evidence` tool, but it is not required.\nHowever, it should be called before the `answer` tool at\nleast one time, but more may be given if specifically helpful for\nthe task, so that the `answer` tool can use the artifacts for the answer.\n\nAn artifact can be any string representable fact or list of facts,\nformatted using markdown syntax. The artifact will be generated as one of\nthe following types:\n- table: Markdown tables with rows and columns (e.g., findings, methods, molecules)\n- code_block: Code snippets in fenced code blocks\n- blockquote: Quoted text or important facts (using > prefix)\n- image: AI-generated image (chart, diagram, or visualization)\n\nEach artifact consists of:\n- A short title (3-8 words) that names what the artifact contains\n- A caption (1-2 sentences) describing what the artifact shows and why it's useful\n- Markdown content in a single element type (validated automatically with retries if needed)\n For image type, content is a markdown image embed ``![title](data:image/png;base64,...)``,\n embedding the PNG directly as a base64 data URL.", + "parameters": { + "type": "object", + "properties": { + "artifact_type": { + "description": "Artifact type to create. Must be one of: table,\ncode_block, blockquote, image.", + "title": "Artifact Type", + "type": "string" + }, + "description": { + "description": "Description of what you want in the artifact. Be specific\nabout what information should be included, how it should be organized,\nand what columns/sections/content you want. This guides the summary LLM.\nFor image type, describe the visual you want generated in detail.", + "title": "Description", + "type": "string" + } + }, + "required": [ + "artifact_type", + "description" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "open_targets_search", + "description": "Search for disease-target associations from OpenTargets database.\n\nAdds new disease-target association data to the state and returns top search results.", + "parameters": { + "type": "object", + "properties": { + "disease_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of disease names to search for.\nUsage:\ndisease_names = [\"Alzheimer's\"] ->\n Returns top targets for Alzheimer's\ndisease_names = [\"Alzheimer's\", \"Diabetes\"], target_names = [\"PSEN1\"] ->\n Filters for only PSEN1-Alzheimer's/Diabetes evidence\ndisease_names = [\"Alzheimer's\"], drug_names = [\"Aducanumab\"] ->\n Filters for Alzheimer's disease-target pairs treated by Aducanumab", + "title": "Disease Names" + }, + "target_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of target names to search for.\nUsage:\ntarget_names = [\"BACE1\"] ->\n Returns top diseases associated with BACE1\ntarget_names = [\"BACE1\", \"APP\"], disease_names = [\"Alzheimer's\"] ->\n Filters for only these target-Alzheimer's pairs\ntarget_names = [\"BACE1\"], drug_names = [\"Aducanumab\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Target Names" + }, + "drug_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of drug names to generate or filter results.\nUsage:\ndrug_names = [\"Aducanumab\"] ->\n Returns all disease-target pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], disease_names = [\"Alzheimer's\"] ->\n Filters for Alzheimer's pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], target_names = [\"BACE1\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Drug Names" + } + }, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "collect_cited_papers_in_evidence", + "description": "Collect papers by traversing the citations of relevant papers to increase the paper count.\n\nThis tool has no effect if called when paper count or relevant papers are zero.\nThis tool will find papers that are likely to lead to more relevant evidence.\nThe papers found have a higher chance of relevance because this tool starts from relevant evidence.\nThis tool's usefulness increases with the amount of gathered evidence as there's more citations to consider.", + "parameters": { + "type": "object", + "properties": {}, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "answer", + "description": "Submit your final answer to the user. You MUST call this tool on your final answer.\nIf you mark all objectives as completed but have not yet called this tool, then\nthe answer will not be submitted. Note that this tool should only be called once, as it\nimmediately terminates the run and submits the answer.\n\nWrite an answer based on the evidence from gather_evidence results in message history.\nIf the context provides insufficient information, provide a partial answer and explain\nwhy it is partial (e.g. the question is currently an open question in the field,\nyou were unable to obtain certain papers, etc.).\n\nFor each part of your answer, indicate which sources most support it via citation IDs at\nthe end of sentences, like (pqac-0000ABCD).\nYou are only allowed to cite available context IDs, indicated in previous messages by \"Context IDs that may be cited:\".\n## Valid citation examples, only use comma/space delimited parentheticals:\n\n- (pqac-0000ABCD, pqac-0000EFGH)\n- (pqac-0000IJKL)\n\n## Invalid citation examples:\n\n- (pqac-0000ABCD and pqac-0000EFGH)\n- (pqac-0000ABCD;pqac-0000EFGH)\n- (pqac-0000ABCD-pqac-0000EFGH)\n- pqac-0000ABCD and pqac-0000EFGH\n- Example's work pqac-0000ABCD\n- (pages pqac-0000ABCD)\n- ![description](pqac-0000ABCD)\n- ()\n\nDo not concatenate citation ids, just use them as-is.\n\nWhen writing your answer, embed artifacts if relevant artifacts have been previously\ncreated by the `create_artifact` tool. To embed an artifact, simply include this in your\nanswer, on its own line and without any additional formatting (i.e. no backticks, code fences,\nquotes, additional tags, etc.).\nAlways provide at least some introductory sentence or paragraph before referring to an artifact,\nso that the artifact is properly contextualized.\nHere is an example reference to an artifact with ID \"artifact-XY\":\n\n\nValid artifact IDs are indicated in previous messages by 'Artifact IDs that may be injected into the answer:'.\n\n\nDouble-check your answer:\n\n1. Evidence Match: Re-read the specific evidence snippets you're citing. Does your\n answer actually match what they say? Don't rely on memory from 20+ messages ago.\n\n2. For Comparative Questions (\"which X has greatest/maximum/highest Y\"):\n - Did you examine ALL options mentioned in the evidence?\n - Did you explicitly compare their values?\n - Example: If evidence mentions \"Pancreatic: 0.92, Liver: 0.85, Lung: 0.78\",\n verify you compared all three, not just two.\n\n3. For Specific Value Questions (\"How many\", \"What is the\", specific terms):\n - Does the exact value you're stating appear in the evidence?\n - Are you extracting from the right context (not noise like page numbers)?\n\n4. Sanity Check: Read your answer. Does it directly address what was asked?\n\nDon't use Markdown image syntax like ![...](...) to reference media in your answer,\nbecause media get automatically linked by the system based on your answer's citations.\n\nWrite in the style of a scientific article, with concise sentences and\ncoherent paragraphs. This answer will be used directly,\nso do not add any extraneous information.", + "parameters": { + "type": "object", + "properties": { + "answer": { + "description": "valid answer, cited with citation IDs.", + "title": "Answer", + "type": "string" + } + }, + "required": [ + "answer" + ] + } + } + } + ], + "messages": [ + { + "role": "user", + "content": "Use the tools to answer the question: What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.\n\nStart by making a plan, and update the plan as you go. Your answer will be returned to the user when all plan objectives are completed. Be thorough and ensure your evidence matches the user's question.", + "info": null + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_kcmzfiVWrTdIWQyBi42VuIon", + "type": "function", + "function": { + "arguments": "{\"plan_json\": \"{\\\"objectives\\\":[{\\\"id\\\":1,\\\"objective\\\":\\\"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\\\",\\\"rationale\\\":\\\"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":2,\\\"objective\\\":\\\"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\\\",\\\"rationale\\\":\\\"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":3,\\\"objective\\\":\\\"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\\\",\\\"rationale\\\":\\\"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":4,\\\"objective\\\":\\\"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\\\",\\\"rationale\\\":\\\"The answer should support a concrete implementation decision, not merely list software.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":5,\\\"objective\\\":\\\"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\\\",\\\"rationale\\\":\\\"A structured matrix will make the multi-package comparison actionable.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":6,\\\"objective\\\":\\\"Call the answer tool with a thorough, evidence-cited technical response.\\\",\\\"rationale\\\":\\\"This is required to submit the final answer after all research objectives are complete.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null}]}\"}", + "name": "create_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Plan registered and formatted correctly, with some modifications:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "create_plan", + "tool_call_id": "call_kcmzfiVWrTdIWQyBi42VuIon", + "info": { + "start_ts": 615.740379951, + "end_ts": 615.744231214 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_VIQzh60P2I1sB79k1x1yJiua", + "type": "function", + "function": { + "arguments": "{\"query\": \"kikuchipy open source Python EBSD software paper orix orientation mapping grain segmentation KAM\", \"min_year\": 2018}", + "name": "paper_search" + } + }, + { + "id": "call_26RNCIR7p5ooimL1oHJKDxPe", + "type": "function", + "function": { + "arguments": "{\"query\": \"DefDAP Python EBSD grain reconstruction texture KAM GND software paper\", \"min_year\": 2018}", + "name": "paper_search" + } + }, + { + "id": "call_TWgXFKMItXYFpWEbgg83FfU9", + "type": "function", + "function": { + "arguments": "{\"query\": \"DREAM.3D software grain segmentation EBSD texture analysis paper simplnx Python\", \"min_year\": 2010}", + "name": "paper_search" + } + }, + { + "id": "call_oB8LA7Kpebnw5E9YNkkGkAve", + "type": "function", + "function": { + "arguments": "{\"query\": \"pyxem orientation mapping EBSD grain segmentation texture analysis software paper\", \"min_year\": 2016}", + "name": "paper_search" + } + }, + { + "id": "call_sXDuxIhQcjT0pGraCjqsLz2r", + "type": "function", + "function": { + "arguments": "{\"query\": \"MTEX texture ODF grain reconstruction KAM GND density EBSD software\", \"min_year\": 2005}", + "name": "paper_search" + } + }, + { + "id": "call_I26KQDTI4oHtsQAWeLYGK9RU", + "type": "function", + "function": { + "arguments": "{\"query\": \"Python EBSD post processing grain reconstruction texture pole figure ODF KAM GND open source\"}", + "name": "paper_search" + } + }, + { + "id": "call_5NDupt5Es4nrgT9SSGt0veGH", + "type": "function", + "function": { + "arguments": "{\"query\": \"pyEBSD open source Python EBSD grain KAM GND\"}", + "name": "paper_search" + } + }, + { + "id": "call_TcPaoKE5klpaxx5d8PlBUDrk", + "type": "function", + "function": { + "arguments": "{\"query\": \"pymicro crystallographic texture EBSD grain Python software\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Machine learning enhanced analysis of EBSD data for texture representation\nValid Text Names:\n'wanni2024machinelearningenhanced pages 1-2', 'wanni2024machinelearningenhanced pages 2-3', 'wanni2024machinelearningenhanced pages 3-5', 'wanni2024machinelearningenhanced pages 5-6', 'wanni2024machinelearningenhanced pages 6-7', 'wanni2024machinelearningenhanced pages 7-9', 'wanni2024machinelearningenhanced pages 9-10', 'wanni2024machinelearningenhanced pages 10-11'\n\nBibTex:\n@article{wanni2024machinelearningenhanced,\n author = \"Wanni, J. and Bronkhorst, C. A. and Thoma, D. J.\",\n title = \"Machine learning enhanced analysis of EBSD data for texture representation\",\n year = \"2024\",\n journal = \"npj Computational Materials\",\n volume = \"10\",\n pages = \"1-11\",\n month = \"Jun\",\n doi = \"10.1038/s41524-024-01324-4\",\n url = \"https://doi.org/10.1038/s41524-024-01324-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n\n\nAbstract:\nGenerating reduced-order, synthetic grain structure datasets that accurately represent the measured grain structure of a material is important for reducing the cost and increasing the accuracy of computational crystal plasticity efforts. This study introduces a machine-learning-based approach, termed texture adaptive clustering and sampling (TACS), for generating representative Euler angle datasets that accurately mimic the crystallographic texture. The TACS approach employs K-means clustering and density-based sampling in a closed-loop iteration to create representative Euler angle datasets. Proof-of-principle experiments were performed on rolled and recrystallized low-carbon steel. Validation of the TACS approach was extended to twenty-two datasets, varying lattice structures, and complex crystallographic textures, thereby encompassing a broad range of materials and crystal structures. Kolmogorov-Smirnov (K-S) test comparisons underscore the performance of the TACS approach over traditional electron backscatter diffraction EBSD dataset reduction techniques, with average K-S test scores nearing 0.9, indicating a high-fidelity representation of the original datasets. In contrast, conventional methods display scores below 0.3, indicating less reliability of the structure representation. The independence of the TACS approach from material texture and its capability to autonomously generate datasets with predetermined data points demonstrates its unbiased potential in streamlining dataset preparation for crystallographic analysis.\n\nRelevant Snippet:\n\u2026 and kernel average misorientation (KAM) maps to deconvolute \u2026 of MTEX, a MATLAB toolbox for texture analysis, the ODF is \u2026 kernel density estimation (KDE), kernel density estimation \u2026\n\n\n---\n\n## 2. Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy\nValid Text Names:\n'bachmann2022efficientreconstructionof pages 1-2', 'bachmann2022efficientreconstructionof pages 2-3', 'bachmann2022efficientreconstructionof pages 3-5', 'bachmann2022efficientreconstructionof pages 5-6', 'bachmann2022efficientreconstructionof pages 6-7', 'bachmann2022efficientreconstructionof pages 7-10', 'bachmann2022efficientreconstructionof pages 10-11', 'bachmann2022efficientreconstructionof pages 11-15', 'bachmann2022efficientreconstructionof pages 15-16', 'bachmann2022efficientreconstructionof pages 16-17', 'bachmann2022efficientreconstructionof pages 17-18', 'bachmann2022efficientreconstructionof pages 18-18'\n\nBibTex:\n@article{bachmann2022efficientreconstructionof,\n author = \"Bachmann, Bj\u00f6rn-Ivo and M\u00fcller, Martin and Britz, Dominik and Durmaz, Ali Riza and Ackermann, Marc and Shchyglo, Oleg and Staudt, Thorsten and M\u00fccklich, Frank\",\n title = \"Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy\",\n year = \"2022\",\n journal = \"Frontiers in Materials\",\n volume = \"9\",\n month = \"Oct\",\n doi = \"10.3389/fmats.2022.1033505\",\n url = \"https://doi.org/10.3389/fmats.2022.1033505\",\n publisher = \"Frontiers Media SA\",\n issn = \"2296-8016\"\n}\n\n\nAbstract:\nThe high-temperature austenite phase is the initial state of practically all technologically relevant hot forming and heat treatment operations in steel processing. The phenomena occurring in austenite, such as recrystallization or grain growth, can have a decisive influence on the subsequent properties of the material. After the hot forming or heat treatment process, however, the austenite transforms into other microstructural constituents and information on the prior austenite morphology are no longer directly accessible. There are established methods available for reconstructing former austenite grain boundaries via metallographic etching or electron backscatter diffraction (EBSD) which both exhibit shortcomings. While etching is often difficult to reproduce and strongly depend on the investigated steel\u2019s alloying concept, EBSD acquisition and reconstruction is rather time-consuming. But in fact, though, light optical micrographs of steels contrasted with conventional Nital etchant also contain information about the former austenite grains. However, relevant features are not directly apparent or accessible with conventional segmentation approaches. This work presents a deep learning (DL) segmentation of prior austenite grains (PAG) from Nital etched light optical micrographs. The basis for successful segmentation is a correlative characterization from EBSD, light and scanning electron microscopy to specify the ground truth required for supervised learning. The DL model shows good and robust segmentation results. While the intersection over union of 70% does not fully reflect the model performance due to the inherent uncertainty in PAG estimation, a mean error of 6.1% in mean grain size derived from the segmentation clearly shows the high quality of the result.\n\nRelevant Snippet:\n\u2026 segmentation is a correlative characterization from EBSD, \u2026 orientation data from electron backscatter diffraction (EBSD) \u2026 This paper represents the first publication of the consortiums\u2019 \u2026\n\n\nUnobtainable Papers:\nDavid J. Rowenhorst, Patrick G. Callahan, and H\u00e5kon W. \u00c5nes. Fast radon transforms for high-precision ebsd orientation determination using pyebsdindex. Journal of Applied Crystallography, 57:3-19, Feb 2024. URL: https://doi.org/10.1107/s1600576723010221, doi:10.1107/s1600576723010221.\nS. Breumier, T. Martinez Ostormujof, B. Frincu, N. Gey, A. Couturier, N. Loukachenko, P.E. Aba-perea, and L. Germain. Leveraging ebsd data by deep learning for bainite, ferrite and martensite segmentation. Materials Characterization, 186:111805, Apr 2022. URL: https://doi.org/10.1016/j.matchar.2022.111805, doi:10.1016/j.matchar.2022.111805.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_VIQzh60P2I1sB79k1x1yJiua", + "info": { + "start_ts": 625.522755479, + "end_ts": 663.10598677 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nIoannis Alakiozidis, Callum Hunt, Albert D. Smith, Mia Maric, Zaheen Shah, A. Ambard, and Philipp Frankel. Microstructure and mechanical performance of cold spray cr coatings. Nov 2024. URL: https://doi.org/10.5281/zenodo.14034890, doi:10.5281/zenodo.14034890.\nPrzemys\u0142aw Snopi\u0144ski and Ond\u0159ej Hil\u0161er. Mechanism of grain refinement in 3d-printed alsi10mg alloy subjected to severe plastic deformation. Materials, 17:4098, Aug 2024. URL: https://doi.org/10.3390/ma17164098, doi:10.3390/ma17164098.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_26RNCIR7p5ooimL1oHJKDxPe", + "info": { + "start_ts": 625.558711009, + "end_ts": 664.725048365 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach\nValid Text Names:\n'yi2025rapidgrainsegmentation pages 1-2', 'yi2025rapidgrainsegmentation pages 2-4', 'yi2025rapidgrainsegmentation pages 4-7', 'yi2025rapidgrainsegmentation pages 7-9', 'yi2025rapidgrainsegmentation pages 9-11', 'yi2025rapidgrainsegmentation pages 11-13', 'yi2025rapidgrainsegmentation pages 13-14'\n\nBibTex:\n@article{yi2025rapidgrainsegmentation,\n author = \"Yi, Yu-Tsen and Seo, Junwon and Murphy, Kevin and Rollett, Anthony D.\",\n title = \"Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach\",\n year = \"2025\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jan\",\n doi = \"10.1007/s40192-024-00390-2\",\n url = \"https://doi.org/10.1007/s40192-024-00390-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\n\n Grain size distribution is a critical factor in determining materials\u2019 physical and mechanical properties, including thermal conductivity, hardness, and creep behavior. Understanding the distribution of grain sizes is essential for advancing the comprehension of material properties and improving materials development and design. Traditional methods for determining grain size, such as electron backscatter diffraction (EBSD), are resource-intensive, underscoring the need for more efficient approaches to grain segmentation in standard micrographs, such as those obtained via SEM and optical imaging. This paper presents a streamlined, unsupervised computer vision pipeline that employs superpixel segmentation and region adjacency merging techniques to segment and measure grain geometry from micrographs efficiently. The pipeline is validated using two methods: hand-labeled SEM images of laser powder bed fusion (LPBF) fabricated Haynes 282 Ni-alloy and open-source EBSD data of IN100 from Dream3D. Both validation approaches achieved IoU and Dice scores greater than 0.9, while processing an image with a resolution of 1000 \u00d7 1000 pixels in under 40\u00a0s, demonstrating a fast and sufficiently accurate pipeline.\n\nRelevant Snippet:\n\u2026 (EBSD) data and Dream3D software to generate grain boundary maps for comparison with the segmentation \u2026 Uchic et al. from the Air Force Research Laboratory (AFRL), was used in \u2026\n\n\n---\n\n## 2. Modeling complex polycrystalline alloys using a Generative Adversarial Network enabled computational platform\nValid Text Names:\n'murgas2024modelingcomplexpolycrystalline pages 1-2', 'murgas2024modelingcomplexpolycrystalline pages 2-3', 'murgas2024modelingcomplexpolycrystalline pages 3-4', 'murgas2024modelingcomplexpolycrystalline pages 4-6', 'murgas2024modelingcomplexpolycrystalline pages 6-7', 'murgas2024modelingcomplexpolycrystalline pages 7-10', 'murgas2024modelingcomplexpolycrystalline pages 10-11', 'murgas2024modelingcomplexpolycrystalline pages 11-12', 'murgas2024modelingcomplexpolycrystalline pages 12-13', 'murgas2024modelingcomplexpolycrystalline pages 13-14', 'murgas2024modelingcomplexpolycrystalline pages 14-15', 'murgas2024modelingcomplexpolycrystalline pages 15-16', 'murgas2024modelingcomplexpolycrystalline pages 16-16'\n\nBibTex:\n@article{murgas2024modelingcomplexpolycrystalline,\n author = \"Murgas, Brayan and Stickel, Joshua and Brewer, Luke and Ghosh, Somnath\",\n title = \"Modeling complex polycrystalline alloys using a Generative Adversarial Network enabled computational platform\",\n year = \"2024\",\n journal = \"Nature Communications\",\n volume = \"15\",\n month = \"Nov\",\n doi = \"10.1038/s41467-024-53865-3\",\n url = \"https://doi.org/10.1038/s41467-024-53865-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2041-1723\"\n}\n\n\nAbstract:\nCreating statistically equivalent virtual microstructures (SEVM) for polycrystalline materials with complex microstructures that encompass multi-modal morphological and crystallographic distributions is a challenging enterprise. Cold spray-formed (CSF) AA7050 alloy containing coarse-grained prior particles and ultra-fine grains (UFG) and additively manufactured (AM) Ti64 alloys with alpha laths in beta substrates. The paper introduces an approach strategically integrating a Generative Adversarial Network (GAN) for multi-modal microstructures with a synthetic microstructure builder DREAM.3D for packing grains conforming to statistics in electron backscatter diffraction (EBSD) maps for generating SEVMs of CSF and AM alloy microstructures. A robust multiscale model is subsequently developed for self-consistent coupling of crystal plasticity finite element model (CPFEM) for coarse-grained crystals with an upscaled constitutive model for UFGs. Sub-volume elements are simulated for efficient computations and their responses are averaged for overall stress-strain response. The methods developed are important for image-based micromechanical modeling that is necessary for microstructure-property relations.\n\nRelevant Snippet:\n\u2026 12e, f, the SEVM values lie between those of the EBSD maps. The crystallographic orientation of every variant is taken from the EBSD data, and the grains are segmented. Figure 5d\u2013f \u2026\n\n\n---\n\n## 3. Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\nValid Text Names:\n'stanek2020comparisonofsegmentation pages 1-2', 'stanek2020comparisonofsegmentation pages 2-7', 'stanek2020comparisonofsegmentation pages 7-9', 'stanek2020comparisonofsegmentation pages 9-12', 'stanek2020comparisonofsegmentation pages 12-16', 'stanek2020comparisonofsegmentation pages 16-18', 'stanek2020comparisonofsegmentation pages 18-19'\n\nBibTex:\n@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n\n\nAbstract:\nA comparison study of grain segmentation in 2D and 3D electron backscatter diffraction (EBSD) measurements of polycrystalline materials, based on a misorientation threshold, is presented. Using Laguerre tessellations with orientation marks, a simulation study that demonstrates a different sensitivity of segmentation in the case of low angle and high angle grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investigated concerning segmentation, and there are given some recommendations for the misorientation thresholds. Computational tools such as DREAM 3D software enable significant data evaluation that is completed by a five-parameter analysis of grain boundaries. K e y w o r d s: electron backscatter diffraction (EBSD), grain boundaries, misorientation, tessellation\n\nRelevant Snippet:\n\u2026 measured by 3D EBSD are investigated concerning segmentation, and \u2026 DREAM 3D software enable significant data evaluation that is completed by a five-parameter analysis of grain \u2026\n\n\n---\n\n## 4. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\nValid Text Names:\n'folwarczny2026accurategrainboundary pages 1-3', 'folwarczny2026accurategrainboundary pages 3-5', 'folwarczny2026accurategrainboundary pages 5-9', 'folwarczny2026accurategrainboundary pages 9-12', 'folwarczny2026accurategrainboundary pages 12-15', 'folwarczny2026accurategrainboundary pages 15-18', 'folwarczny2026accurategrainboundary pages 18-20'\n\nBibTex:\n@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n\n\nRelevant Snippet:\nWe present a method for obtaining qualitatively accurate grain boundary plane distributions (GBPD) for textured microstructures using a stereological calculation applied to two-\u2026\n\n\nUnobtainable Papers:\nRandi Swanson, Michael Chapman, Yue Zhou, Ashley Hilmas, Lisa Rueschhoff, Michael Uchic, William G. Fahrenholtz, and Scott J. McCormack. Quantitative grain structure and texture analysis of hot-pressed zrb2 via 3d ebsd. Materials & Design, 261:115378, Jan 2026. URL: https://doi.org/10.1016/j.matdes.2025.115378, doi:10.1016/j.matdes.2025.115378.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_TWgXFKMItXYFpWEbgg83FfU9", + "info": { + "start_ts": 625.591091176, + "end_ts": 664.734003185 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Automated Crystal Orientation Mapping in py4DSTEM using Sparse Correlation Matching\nValid Text Names:\n'ophus2022automatedcrystalorientation pages 1-2', 'ophus2022automatedcrystalorientation pages 2-4', 'ophus2022automatedcrystalorientation pages 4-5', 'ophus2022automatedcrystalorientation pages 5-6', 'ophus2022automatedcrystalorientation pages 6-7', 'ophus2022automatedcrystalorientation pages 7-8', 'ophus2022automatedcrystalorientation pages 8-10', 'ophus2022automatedcrystalorientation pages 10-11', 'ophus2022automatedcrystalorientation pages 11-12', 'ophus2022automatedcrystalorientation pages 12-13', 'ophus2022automatedcrystalorientation pages 13-14'\n\nBibTex:\n@article{ophus2022automatedcrystalorientation,\n author = \"Ophus, Colin and Zeltmann, Steven E and Bruefach, Alexandra and Rakowski, Alexander and Savitzky, Benjamin H and Minor, Andrew M and Scott, MC\",\n title = \"Automated Crystal Orientation Mapping in py4DSTEM using Sparse Correlation Matching\",\n year = \"2022\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"28\",\n pages = \"390-403\",\n month = \"Oct\",\n doi = \"10.48550/arxiv.2111.00171\",\n url = \"https://doi.org/10.48550/arxiv.2111.00171\",\n publisher = \"arXiv\"\n}\n\n\nAbstract:\nAbstract Crystalline materials used in technological applications are often complex assemblies composed of multiple phases and differently oriented grains. Robust identification of the phases and orientation relationships from these samples is crucial, but the information extracted from the diffraction condition probed by an electron beam is often incomplete. We have developed an automated crystal orientation mapping (ACOM) procedure which uses a converged electron probe to collect diffraction patterns from multiple locations across a complex sample. We provide an algorithm to determine the orientation of each diffraction pattern based on a fast sparse correlation method. We demonstrate the speed and accuracy of our method by indexing diffraction patterns generated using both kinematical and dynamical simulations. We have also measured orientation maps from an experimental dataset consisting of a complex polycrystalline twisted helical AuAgPd nanowire. From these maps we identify twin planes between adjacent grains, which may be responsible for the twisted helical structure. All of our methods are made freely available as open source code, including tutorials which can be easily adapted to perform ACOM measurements on diffraction pattern datasets.\n\nRelevant Snippet:\n\u2026 electron backscatter diffraction (EBSD) in scanning electron \u2026 disks in the pattern, then segmenting the data into radial bands\u2026 the orientation mapping application considered in this \u2026\n\n\n---\n\n## 2. High resolution crystal orientation mapping of ultrathin films in SEM and TEM\nValid Text Names:\n'heinig2022highresolutioncrystal pages 1-2', 'heinig2022highresolutioncrystal pages 2-3', 'heinig2022highresolutioncrystal pages 3-5', 'heinig2022highresolutioncrystal pages 5-6', 'heinig2022highresolutioncrystal pages 6-7', 'heinig2022highresolutioncrystal pages 7-8', 'heinig2022highresolutioncrystal pages 8-9', 'heinig2022highresolutioncrystal pages 9-9'\n\nBibTex:\n@misc{heinig2022highresolutioncrystal,\n author = \"Heinig, Mario F. and Chatterjee, Dipanwita and van Helvoort, Antonius T.J. and Wagner, Jakob Birkedal and Kadkhodazadeh, Shima and \u00c5nes, H\u00e5kon Wiik and Niessen, Frank and da Silva Fanta, Alice Bastos\",\n title = \"High resolution crystal orientation mapping of ultrathin films in SEM and TEM\",\n year = \"2022\",\n journal = \"Materials Characterization\",\n volume = \"189\",\n pages = \"111931\",\n month = \"Jul\",\n doi = \"10.1016/j.matchar.2022.111931\",\n url = \"https://doi.org/10.1016/j.matchar.2022.111931\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n\n\nRelevant Snippet:\nUltrathin metallic films are important functional materials for optical and microelectronic devices. Dedicated characterization with high spatial resolution and sufficient field of view is key \u2026\n\n\nUnobtainable Papers:\nDavid P. Field, Satyam Suwas, and Stuart I. Wright. Basics of SEM-Based Techniques for Electron Backscatter Diffraction, pages 27-41. Springer Nature Singapore, Jan 2025. URL: https://doi.org/10.1007/978-981-96-5346-1\\_2, doi:10.1007/978-981-96-5346-1\\_2.\nMarco Andres Acevedo Zamora, Christoph Eckart Schrank, and Balz Samuel Kamber. Using the traditional microscope for mineral grain orientation determination: a prototype image analysis pipeline for optic\u2010axis mapping (poam). Journal of Microscopy, 295:147-176, Mar 2024. URL: https://doi.org/10.1111/jmi.13284, doi:10.1111/jmi.13284.\nThomas Girerd, Giovanna Mart\u00ednez-Arellano, Adam Thomas Clare, and Alistair Speidel. A facile methodology to identify microstructural grains on etched surfaces using panoptic segmentation. Sep 2025. URL: https://doi.org/10.1016/j.matchar.2025.115224, doi:10.1016/j.matchar.2025.115224.\nYukun Liu, Patricia E. Meza, Zhi Li, Xiaobing Hu, Roberto dos Reis, Christopher Wolverton, and Vinayak Dravid. Seeing the invisible-understanding thermoelectric materials through scanning transmission electron microscopy. Chemistry of Materials, 38(8):3855-3878, Apr 2026. URL: https://doi.org/10.1021/acs.chemmater.5c02947, doi:10.1021/acs.chemmater.5c02947.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_oB8LA7Kpebnw5E9YNkkGkAve", + "info": { + "start_ts": 625.626175119, + "end_ts": 663.770777395 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nQiang Zhu, Chuanjie Wang, He-yong Qin, Gang Chen, and Peng Zhang. Effect of the grain size on the microtensile deformation and fracture behaviors of a nickel-based superalloy via ebsd and in-situ synchrotron radiation x-ray tomography. Materials Characterization, 156:109875, Oct 2019. URL: https://doi.org/10.1016/j.matchar.2019.109875, doi:10.1016/j.matchar.2019.109875.\nHao Chen, Yang Tao, Cheng Xin Li, and Han Xiao. The effect of grain size on the microstructural deformation behavior of cusn10p1 alloy under quasi-in-situ compression. Materials Science and Engineering: A, 973:150763, Oct 2026. URL: https://doi.org/10.1016/j.msea.2026.150763, doi:10.1016/j.msea.2026.150763.\nYaozhong Zhang, James Ball, Axel Henningsson, Jon Wright, Lucero Lopez, Dillon Jobes, Henry Proudhon, and Jerard V. Gordon. Unveiling 3d sub-grain residual stresses in as-built additively manufactured steel using scanning 3dxrd. Materials Research Letters, 13:700-708, May 2025. URL: https://doi.org/10.1080/21663831.2025.2502502, doi:10.1080/21663831.2025.2502502.\nMartin Diehl and Markus K\u00fchbach. Coupled experimental-computational analysis of primary static recrystallization in low carbon steel. Modelling and Simulation in Materials Science and Engineering, 28:014001, Nov 2019. URL: https://doi.org/10.1088/1361-651x/ab51bd, doi:10.1088/1361-651x/ab51bd.\nSven E Gustafson. Sub-grain deformation mechanisms of structural materials exposed through x-ray microscopy. Text, Jan 2023. URL: https://doi.org/10.25394/pgs.22151402, doi:10.25394/pgs.22151402.\njeremie bouquerel and Jean-Bernard VOGT. Assessment of ebsd regarding cyclic plasticity: a point of view. Unknown journal, Jan 2024. URL: https://doi.org/10.2139/ssrn.5077433, doi:10.2139/ssrn.5077433.\nLeo A.I. Kestens, Jesus Galan Lopez, Hadi Pirgazi, Tuan Nguyen-Minh, and Roumen H. Petrov. Texture observation and control in metal manufacturing : theory and practice. IOP Conference Series: Materials Science and Engineering, 1121:012002, Mar 2021. URL: https://doi.org/10.1088/1757-899x/1121/1/012002, doi:10.1088/1757-899x/1121/1/012002.\nZhijun Qiu, Hanliang Zhu, Zhiyang Wang, Ondrej Mur\u00e1nsky, Tao Wei, Elizabeth Budzakoska-Testone, Joel Davis, Andrew Studer, Q. F. Gu, Ulf Garbe, David Wexler, and Huijun Li. Crystallographic texture and multiscale boundaries mediated creep anisotropy in additively manufactured ni-based hastelloy c276 superalloy. Additive Manufacturing, Mar 2024. URL: https://doi.org/10.1016/j.addma.2024.104069, doi:10.1016/j.addma.2024.104069.\nSrinivas Doddapaneni, Sathish Kumar, Sathyashankara Sharma, Gowri Shankar, Manjunath Shettar, Nitesh Kumar, Ganesha Aroor, and Syed Mansoor Ahmad. Advancements in ebsd techniques: a comprehensive review on characterization of composites and metals, sample preparation, and operational parameters. Journal of Composites Science, 9:132, Mar 2025. URL: https://doi.org/10.3390/jcs9030132, doi:10.3390/jcs9030132.\nSoroosh Naghdy, L. L. Percq, R. Serret, Roumen Petrov, S. Hertel\u00e9, L. Kestens, and P. Verleysen. Microstructural evolution study of severely deformed commercial aluminium by transmission kikuchi diffraction. Materials Science and Technology, 33:678-687, Apr 2017. URL: https://doi.org/10.1080/02670836.2016.1194019, doi:10.1080/02670836.2016.1194019.\nF. Cruz-Gandarilla, \u2020. R.E.BOLMARO, \u2021. H.F.MENDOZA-LE\u00b4ON, A. M. S. J. C. \u02dcNAS-MORENO, F. Cruz-Gandarilla, Escuela Superior F\u00b4\u0131sica, Matem\u00b4aticas Instituto, Edificio Polit\u00b4ecnico Nacional, Del. G.A Madero, and C.P.07738. Ciudad de M\u00b4exico. Study of recovery and first recrystallisation kinetics in cgo fe3%si steels using misorientation\u2010derived parameters (ebsd). Journal of Microscopy, 275:133-148, Aug 2019. URL: https://doi.org/10.1111/jmi.12822, doi:10.1111/jmi.12822.\nJ. Collins, M. Taylor, A.L. Scarlett, E.J. Palmiere, and E.J. Pickering. Prior austenite grain measurement: a direct comparison of ebsd reconstruction, thermal etching and chemical etching. Materials Characterization, 208:113656, Feb 2024. URL: https://doi.org/10.1016/j.matchar.2024.113656, doi:10.1016/j.matchar.2024.113656.\n\u0160t\u011bp\u00e1n Gamanov, Poczkl\u00e1n Ladislav, Antonin Dlouhy, and Tomas Kruml. A step size dependence of kam, grod and gnd parameters calculated from ebsd. Jan 2025. URL: https://doi.org/10.2139/ssrn.4789333, doi:10.2139/ssrn.4789333.\nMoritz M\u00fcller, Amaya Compart, Mahnaz Mehdizadehlima, Serhii Yaroshevskyi, Christos G. Aneziris, Horst Biermann, and Anja Weidner. Ebsd study of creep\u2010induced lattice misorientation in mgo\u2010particle\u2010reinforced austenitic steel composites. Advanced Engineering Materials, Mar 2026. URL: https://doi.org/10.1002/adem.202502523, doi:10.1002/adem.202502523.\nFulin Wang, Jean-Charles Stinville, Marie Charpagne, McLean P. Echlin, Sean R. Agnew, Tresa M. Pollock, Marc De Graef, and Daniel S. Gianola. Dislocation cells in additively manufactured metallic alloys characterized by electron backscatter diffraction pattern sharpness. Mar 2023. URL: https://doi.org/10.1016/j.matchar.2023.112673, doi:10.1016/j.matchar.2023.112673.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_sXDuxIhQcjT0pGraCjqsLz2r", + "info": { + "start_ts": 625.665071324, + "end_ts": 664.716058467 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nJingwen Gao, Chenyang Zhu, Shubo Gao, Weiming Ji, Ming Xue, and Kun Zhou. Generative adversarial network\u2013enabled microstructural mapping from surface profiles for laser powder bed fusion. Virtual and Physical Prototyping, Jun 2025. URL: https://doi.org/10.1080/17452759.2025.2499925, doi:10.1080/17452759.2025.2499925.\nkikuchipy: an open-source toolbox for analysis of EBSD patterns\nJie Jiang, Xiaochen Li, and Zhenbo Zhang. Effect of scanning strategy on the formation of heterostructure and recrystallization behavior of additively manufactured in718 alloy. Jul 2026. URL: https://doi.org/10.1016/j.jmrt.2026.06.138, doi:10.1016/j.jmrt.2026.06.138.\nM. Fekih, T. Martinez Ostormujof, L. Germain, M. Piette, D. Sornin, and N. Gey. High temperature ebsd experiment versus crystallographic reconstruction to study phase transformation induced microstructures - application examples for steels. Materials Characterization, 228:115427, Oct 2025. URL: https://doi.org/10.1016/j.matchar.2025.115427, doi:10.1016/j.matchar.2025.115427.\nJinghua Cong, Shengjie Yang, Pengcheng Liu, Haibin Zhu, Xueliang Shang, Yongzhe Yang, and Xuemin Wang. Gradient temperature rolling induced homogenization of through-thickness mechanical properties and improved surface corrosion resistance in a heavy steel plate. Journal of Materials Research and Technology, Jun 2026. URL: https://doi.org/10.1016/j.jmrt.2026.06.009, doi:10.1016/j.jmrt.2026.06.009.\nIn Situ High Energy X-ray Diffraction Characterization of Phase Transformations and Mechanical Behaviors in Rapidly Solidified Titanium and Stainless Steel Alloys\nW.T. Zhang, R. Jiang, Y. Zhao, L.C. Zhang, L. Zhang, L.G. Zhao, and Y.D. Song. Effects of temperature and microstructure on low cycle fatigue behaviour of a pm ni-based superalloy: ebsd assessment and crystal plasticity simulation. International Journal of Fatigue, 159:106818, Jun 2022. URL: https://doi.org/10.1016/j.ijfatigue.2022.106818, doi:10.1016/j.ijfatigue.2022.106818.\nRong-lei Fan, Yong Wu, Minghe Chen, and Jie Zhao. Prediction of anisotropic deformation behavior of ta32 titanium alloy sheet during hot tension by crystal plasticity finite element model. May 2022. URL: https://doi.org/10.1016/j.msea.2022.143137, doi:10.1016/j.msea.2022.143137.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_I26KQDTI4oHtsQAWeLYGK9RU", + "info": { + "start_ts": 625.702459775, + "end_ts": 667.448846894 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures\nValid Text Names:\n'ostormujof2022deeplearningfor pages 1-3', 'ostormujof2022deeplearningfor pages 3-5', 'ostormujof2022deeplearningfor pages 5-9', 'ostormujof2022deeplearningfor pages 9-13', 'ostormujof2022deeplearningfor pages 13-16', 'ostormujof2022deeplearningfor pages 16-18', 'ostormujof2022deeplearningfor pages 18-21'\n\nBibTex:\n@article{ostormujof2022deeplearningfor,\n author = \"Ostormujof, T Martinez and Purohit, Rrp Purushottam Raj and Breumier, S and Gey, Nathalie and Salib, M and Germain, L\",\n title = \"Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures\",\n year = \"2022\",\n journal = \"Materials Characterization\",\n month = \"Nov\",\n doi = \"10.48550/arxiv.2112.03072\",\n url = \"https://doi.org/10.48550/arxiv.2112.03072\",\n publisher = \"arXiv\"\n}\n\n\nRelevant Snippet:\n\u2026 ) are able to extract meaningful features from EBSD-based data in order to automatically \u2026 parameters such as Kernel Average Misorientation (KAM) or pattern quality. Compared to other \u2026\n\n\n---\n\n## 2. Generalizable image segmentation for microstructure characterization through integrated SEM and EBSD analysis\nValid Text Names:\n'taufique2025generalizableimagesegmentation pages 1-2', 'taufique2025generalizableimagesegmentation pages 2-3', 'taufique2025generalizableimagesegmentation pages 3-4', 'taufique2025generalizableimagesegmentation pages 4-6', 'taufique2025generalizableimagesegmentation pages 6-7', 'taufique2025generalizableimagesegmentation pages 7-8', 'taufique2025generalizableimagesegmentation pages 8-10', 'taufique2025generalizableimagesegmentation pages 10-10'\n\nBibTex:\n@article{taufique2025generalizableimagesegmentation,\n author = \"Taufique, M. F. N. and Nguyen, Julia and Escobar, Julian D. and Pole, Mayur and Garcia, David and Wang, Tianhao and Das, Hrishikesh and Nwe, Kathy and Brown, David and Ross, Kenneth and Todd, Donald R. and Barker, Erin and Smith, Eric and Kappagantula, Keerti S. and Bilbrey, Jenna A.\",\n title = \"Generalizable image segmentation for microstructure characterization through integrated SEM and EBSD analysis\",\n year = \"2025\",\n journal = \"npj Computational Materials\",\n volume = \"11\",\n month = \"Oct\",\n doi = \"10.1038/s41524-025-01801-4\",\n url = \"https://doi.org/10.1038/s41524-025-01801-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n\n\nAbstract:\nAbstract\n We demonstrate generalizable semantic segmentation using minimal ground truth data. Correlated scanning electron microscopy (SEM) images and electron backscatter diffraction (EBSD) measurements of friction-stir processed 316L stainless steel plates were used to train deep learning models for grain boundary segmentation. Secondary electron (SE) imaging taken at 10\u2009keV correlated to EBSD-derived grain boundaries produced the best performing model. Notably, an ensemble of three models trained on a single SE image produced accurate segmentation over a series of backscatter electron (BSE) images of samples manufactured under different processing parameters, with a mean absolute error in grain size of 0.34\u2009\u00b5m. The generalizability of the models likely results from the similar escape depths of the SE training input and the EBSD training output and the reduced probability of stored strain artifacts appearing in the image. This highlights the importance of considering the physical principles behind imaging to develop robust models for microstructure characterization.\n\nRelevant Snippet:\n\u2026 seen by EBSD, which alters our perception of the ground truth \u2026 of the EBSD data, we reconstructed the grain boundaries \u2026 via kernel average misorientation (KAM) analysis. \u2026\n\n\nUnobtainable Papers:\nFederico Bruno, Georgios Konstantopoulos, Gianluca Fiore, Edoardo Rossi, Marco Sebastiani, Costas Charitidis, Luca Belforte, and Mauro Palumbo. A novel machine learning method to exploit ebsd and nanoindentation for trip steels microstructures analysis. Mar 2024. URL: https://doi.org/10.1016/j.matdes.2024.112774, doi:10.1016/j.matdes.2024.112774.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_5NDupt5Es4nrgT9SSGt0veGH", + "info": { + "start_ts": 625.742518522, + "end_ts": 662.189370401 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes\nValid Text Names:\n'polonsky2026pytribeamopensourcesoftware pages 1-2', 'polonsky2026pytribeamopensourcesoftware pages 2-5', 'polonsky2026pytribeamopensourcesoftware pages 5-7', 'polonsky2026pytribeamopensourcesoftware pages 7-8', 'polonsky2026pytribeamopensourcesoftware pages 8-10', 'polonsky2026pytribeamopensourcesoftware pages 10-12', 'polonsky2026pytribeamopensourcesoftware pages 12-14', 'polonsky2026pytribeamopensourcesoftware pages 14-15', 'polonsky2026pytribeamopensourcesoftware pages 15-16', 'polonsky2026pytribeamopensourcesoftware pages 16-17'\n\nBibTex:\n@article{polonsky2026pytribeamopensourcesoftware,\n author = \"Polonsky, Andrew T. and Lamb, James D. and Hovey, Chad B. and Schroader, Haydn and Echlin, McLean P. and Pollock, Tresa M.\",\n title = \"pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes\",\n year = \"2026\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"15\",\n pages = \"1-17\",\n month = \"Feb\",\n doi = \"10.1007/s40192-026-00449-2\",\n url = \"https://doi.org/10.1007/s40192-026-00449-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 the RAMS systems [15], and PyMicro based in situ testing [16]\u2026 Scientific (TFS) now have python APIs for microscope control, \u2026 , all EBSD parameters must be filled out in the EBSD vendor \u2026\n\n\n---\n\n## 2. Registration between DCT and EBSD datasets for multiphase microstructures\nValid Text Names:\n'ball2023registrationbetweendct pages 1-2', 'ball2023registrationbetweendct pages 2-4', 'ball2023registrationbetweendct pages 4-6', 'ball2023registrationbetweendct pages 6-7', 'ball2023registrationbetweendct pages 7-9', 'ball2023registrationbetweendct pages 9-11', 'ball2023registrationbetweendct pages 11-12', 'ball2023registrationbetweendct pages 12-14', 'ball2023registrationbetweendct pages 14-15', 'ball2023registrationbetweendct pages 15-15'\n\nBibTex:\n@article{ball2023registrationbetweendct,\n author = \"Ball, J. and Oddershede, J. and Davis, C. and Slater, C. and Said, M. and Vashishtha, H. and Michalik, \u0160. and of Metallurgy, David M. Collins School and Materials and Birmingham, U. and Kingdom., United and Ltd., Diamond Light Source and ApS, Xnovo Technology and Denmark and Wmg and Warwick, U.\",\n title = \"Registration between DCT and EBSD datasets for multiphase microstructures\",\n year = \"2023\",\n journal = \"Materials Characterization\",\n volume = \"204\",\n pages = \"113228\",\n month = \"Oct\",\n doi = \"10.1016/j.matchar.2023.113228\",\n url = \"https://doi.org/10.1016/j.matchar.2023.113228\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n\n\nRelevant Snippet:\n\u2026 the large grains for each phase present, austenite and martensitic ferrite. The results reveals a complex ferrite grain network of similar crystal orientations that are absent from the EBSD \u2026\n\n\nUnobtainable Papers:\nJames A. D. Ball, Anna Kareer, Oxana V. Magdysyuk, Stefan Michalik, Thomas Connolley, and David M. Collins. Per-grain and neighbourhood stress interactions during deformation of a ferritic steel obtained using three-dimensional x-ray diffraction. Text, Jan 2023. URL: https://doi.org/10.48550/arxiv.2302.10711, doi:10.48550/arxiv.2302.10711.\nYanling Schneider, Vighnesh Prabhu, Kai H\u00f6ss, Werner Wasserb\u00e4ch, Siegfried Schmauder, and Zhangjian Zhou. Many-scale investigations of the deformation behavior of polycrystalline composites: i\u2014machine learning applied for image segmentation. Materials, 15:2486, Mar 2022. URL: https://doi.org/10.3390/ma15072486, doi:10.3390/ma15072486.\nSangwon Lee, Michael Pilipchuk, C. Yildirim, Duncan A. Greeley, Qianying Shi, Tracy Berman, Adam Creuziger, Evan Rust, Carsten Detlefs, V. Sundararaghavan, John E. Allison, and Ashley Bucsek. Three-dimensional nucleation and growth of deformation twins in magnesium. Science, 389 6760:632-636, Dec 2025. URL: https://doi.org/10.48550/arxiv.2412.16640, doi:10.48550/arxiv.2412.16640.\nChristopher S. Daniel, Xiaohan Zeng, \u0160tefan Michalik, Simon A. Hunt, and Jo\u00e3o Quinta da Fonseca. Application of a new method for accurate determination of \u03b1 and \u03b2 texture in ti-6al-4v from synchrotron diffraction intensities. May 2023. URL: https://doi.org/10.1016/j.matchar.2023.112769, doi:10.1016/j.matchar.2023.112769.\nPierre Belamri, Henry Proudhon, Damien Texier, and David Ryckelynck. Quaternion-based vision-transformer for polycrystalline ebsd scans pre-trained on large-scale synthetic data. Materials & Design, 258:114599, Oct 2025. URL: https://doi.org/10.1016/j.matdes.2025.114599, doi:10.1016/j.matdes.2025.114599.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_TcPaoKE5klpaxx5d8PlBUDrk", + "info": { + "start_ts": 625.798862717, + "end_ts": 671.189819695 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_YCK1KFOsT4kHD9IBqs5QYRHi", + "type": "function", + "function": { + "arguments": "{\"question\": \"Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "No relevant evidence found for your question.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_YCK1KFOsT4kHD9IBqs5QYRHi", + "info": { + "start_ts": 681.740740548, + "end_ts": 688.630316028 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_YAEYGtulkKwxKg8CZdFRkuKJ", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_6EDjDZUx9EcloPKavvbsh1xb", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"orix\\\" Python crystallographic orientation library\"}", + "name": "paper_search" + } + }, + { + "id": "call_bd44zev0ajEnct80hGQyaouJ", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"DefDAP\\\" deformation data analysis Python\"}", + "name": "paper_search" + } + }, + { + "id": "call_Z5I4Dszqv8IduAWdkGxzojFr", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"DREAM.3D\\\" \\\"Integrating Materials and Manufacturing Innovation\\\" software\"}", + "name": "paper_search" + } + }, + { + "id": "call_5hiGgVfNiBeJ5xEkpggCPT0O", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"MTEX\\\" quantitative texture analysis software\"}", + "name": "paper_search" + } + }, + { + "id": "call_t3X9GkiKh5YtIwvvFGFtqFju", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"PyMicro\\\" Python microstructure library crystallography\"}", + "name": "paper_search" + } + }, + { + "id": "call_2WFpaGkkorc7QNUjgoGtFJOi", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"pyxem\\\" open-source Python library diffraction microscopy\"}", + "name": "paper_search" + } + }, + { + "id": "call_xqX7V5hmrKT4tjzjK7lLSxlE", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"OpenXY\\\" EBSD Python GND\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_YAEYGtulkKwxKg8CZdFRkuKJ", + "info": { + "start_ts": 703.386594257, + "end_ts": 745.729313749 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nD. Johnstone, Ben H. Martineau, P. Crout, P. Midgley, and A. Eggeman. Density-based clustering of crystal (mis)orientations and the orix python library. Journal of Applied Crystallography, 53:1293-1298, Sep 2020. URL: https://doi.org/10.1107/s1600576720011103, doi:10.1107/s1600576720011103.\nIan MacLaren, Enrique Frutos\u2010Myro, Steven Zeltmann, and Colin Ophus. A method for crystallographic mapping of an alpha\u2010beta titanium alloy with nanometre resolution using scanning precession electron diffraction and open\u2010source software libraries. Journal of Microscopy, 295:131-139, Feb 2024. URL: https://doi.org/10.1111/jmi.13275, doi:10.1111/jmi.13275.\nZhuo Li, Daren Wu, Simon Si Ming Ji, Ana G. Claus, Hui Zhong, Sanjit K. Ghose, and Kelsey B. Hatzell. Texture evolution of plated lithium in anode-free solid-state batteries. ACS Energy Letters, 11(2):1705-1713, Jan 2026. URL: https://doi.org/10.1021/acsenergylett.5c03205, doi:10.1021/acsenergylett.5c03205.\nDorian Depriester and R\u00e9gis Kubler. Elasticipy: a python package for linear elasticity and tensor analysis. Journal of Open Source Software, 10:7940, Nov 2025. URL: https://doi.org/10.21105/joss.07940, doi:10.21105/joss.07940.\nMattis Seehaus, Sang-Hyeok Lee, Tobias Stollenwerk, Jeffrey M. Wheeler, and Sandra Korte-Kerzel. Estimation of directional single crystal elastic properties from nano-indentation by correlation with ebsd and first-principle calculations. Materials & Design, 234:112296, Oct 2023. URL: https://doi.org/10.1016/j.matdes.2023.112296, doi:10.1016/j.matdes.2023.112296.\nZachary Varley. Algorithms for crystallography in the scanning electron microscope. Text, Jan 2024. URL: https://doi.org/10.1184/r1/26955991, doi:10.1184/r1/26955991.\nColin Ophus. Quantitative scanning transmission electron microscopy for materials science: imaging, diffraction, spectroscopy, and tomography. Annual Review of Materials Research, 53:105-141, Jul 2023. URL: https://doi.org/10.1146/annurev-matsci-080921-092646, doi:10.1146/annurev-matsci-080921-092646.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_6EDjDZUx9EcloPKavvbsh1xb", + "info": { + "start_ts": 703.425128648, + "end_ts": 746.329071205 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nBenjamin Poole, Alex Marsh, David Lunt, Chris Hardie, Mike Gorley, Cory Hamelin, and Allan Harte. High\u2010resolution strain mapping in a thermionic lab6 scanning electron microscope. Strain, Feb 2024. URL: https://doi.org/10.1111/str.12472, doi:10.1111/str.12472.\nYangchao Deng, Zhenjie Cao, Dongchen Hu, Albert D. Smith, Richard Hunt, Kun Yan, and Joseph D. Robson. Insights into strain partitioning and void nucleation in recycled aluminium alloy: the role of fe-rich intermetallic. Materials Science and Engineering: A, 973:150757, Oct 2026. URL: https://doi.org/10.1016/j.msea.2026.150757, doi:10.1016/j.msea.2026.150757.\nConghui Liu, Xu Xu, Tianzhu Sun, Rhys Thomas, Jo\u00e3o Quinta da Fonseca, and Michael Preuss. Microstructural effects on fatigue crack initiation mechanisms in a near-alpha titanium alloy. Jul 2023. URL: https://doi.org/10.1016/j.actamat.2023.118957, doi:10.1016/j.actamat.2023.118957.\nRhys Thomas, David Bowden, David Lunt, David Stewart, and Michael Preuss. High compressive loading performance of a complex multi-phase hard-facing alloy explained through a highly elastic silicide phase. Aug 2023. URL: https://doi.org/10.1016/j.mtla.2023.101815, doi:10.1016/j.mtla.2023.101815.\nMichael D Atkinson, Rhys Thomas, Allan Harte, Peter Crowther, and Jo\u00e3o Quinta da Fonseca. Defdap: deformation data analysis in python. Feb 2020. URL: https://doi.org/10.5281/zenodo.3688097, doi:10.5281/zenodo.3688097.\nConghui Liu, Tianzhu Sun, Dongchen Hu, Rhys Thomas, Philipp Frankel, Jo\u00e3o Quinta da Fonseca, and Michael Preuss. Mechanistic insight into cooperative slip system activation under cyclic loading in a near-alpha titanium alloy. Acta Materialia, 308:122031, Apr 2026. URL: https://doi.org/10.1016/j.actamat.2026.122031, doi:10.1016/j.actamat.2026.122031.\nRan Ni, Carl J. Boehlert, Bo Chen, Yuanshuai Zhang, Ying Zeng, J. Zheng, Hao Zhou, Qudong Wang, and Dongdi Yin. Temperature-dependent interplay of intra- and inter-granular deformation mechanisms in mg-10y: statistical analysis from an hrdic perspective. Acta Materialia, Jun 2025. URL: https://doi.org/10.1016/j.actamat.2025.121256, doi:10.1016/j.actamat.2025.121256.\nSunil Anandatheertha, Vikram Phalke, Eralp Demir, and Christopher Hardie. Ukaea poly-xtal operations (upxo v1.0): an open-source python package for generating, assessment and meshing of poly-crystalline grain structures. SoftwareX, 34:102736, Jun 2026. URL: https://doi.org/10.1016/j.softx.2026.102736, doi:10.1016/j.softx.2026.102736.\nDongchen Hu, Albert D. Smith, David Lunt, Rhys Thomas, Michael D. Atkinson, Xiaodong Liu, \u00d6mer Ko\u00e7, Jack M. Donoghue, Zhenbo Zhang, Jo\u00e3o Quinta da Fonseca, and Michael Preuss. Tracking the onset of plasticity in a ni-base superalloy using in-situ high-resolution digital image correlation. Feb 2025. URL: https://doi.org/10.1016/j.matchar.2024.114654, doi:10.1016/j.matchar.2024.114654.\nBerzah Yavuzyegit, Egemen Avcu, Albert D. Smith, Jack M. Donoghue, David Lunt, Joseph D. Robson, Timothy L. Burnett, Jo\u00e3o Quinta da Fonseca, and Philip J. Withers. Mapping plastic deformation mechanisms in az31 magnesium alloy at the nanoscale. SSRN Electronic Journal, 250:118876, May 2023. URL: https://doi.org/10.2139/ssrn.4198880, doi:10.2139/ssrn.4198880.\nJie Luo, Albert D. Smith, Jack M. Donoghue, Jiaqi Xu, Juan Nunez-Iglesias, Christopher Hutchinson, Philip J. Withers, and Michael Preuss. Automated in-situ testing of deformation and phase transformations in engineering alloys. Automated In-situ Testing of Deformation and Phase transformations in Engineering Alloys, Jan 2025. URL: https://doi.org/10.14293/apmc13-2025-0293, doi:10.14293/apmc13-2025-0293.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_bd44zev0ajEnct80hGQyaouJ", + "info": { + "start_ts": 703.459767095, + "end_ts": 740.862397661 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\nValid Text Names:\n'polonsky2019threedimensionalanalysisand pages 1-2', 'polonsky2019threedimensionalanalysisand pages 2-5', 'polonsky2019threedimensionalanalysisand pages 5-6', 'polonsky2019threedimensionalanalysisand pages 6-8', 'polonsky2019threedimensionalanalysisand pages 8-11', 'polonsky2019threedimensionalanalysisand pages 11-13', 'polonsky2019threedimensionalanalysisand pages 13-14', 'polonsky2019threedimensionalanalysisand pages 14-15', 'polonsky2019threedimensionalanalysisand pages 15-15'\n\nBibTex:\n@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a cloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline variables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore, a comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using BisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects for impact in material science are also discussed.\n\nRelevant Snippet:\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a cloud-based platform, BisQue. Parallelization of DREAM.3D module \u2026\n\n\n---\n\n## 2. ITKMontage: A Software Module for Image Stitching\nValid Text Names:\n'zukic2021itkmontageasoftware pages 1-2', 'zukic2021itkmontageasoftware pages 2-4', 'zukic2021itkmontageasoftware pages 4-5', 'zukic2021itkmontageasoftware pages 5-8', 'zukic2021itkmontageasoftware pages 8-10', 'zukic2021itkmontageasoftware pages 10-10'\n\nBibTex:\n@article{zukic2021itkmontageasoftware,\n author = \"Zuki\u0107, D\u017eenan and Jackson, Michael and Dimiduk, Dennis and Donegan, Sean and Groeber, Michael and McCormick, Matthew\",\n title = \"ITKMontage: A Software Module for Image Stitching\",\n year = \"2021\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"10\",\n pages = \"115-124\",\n month = \"Mar\",\n doi = \"10.1007/s40192-021-00202-x\",\n url = \"https://doi.org/10.1007/s40192-021-00202-x\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 Its liberal Apache 2.0 license allows incorporation of the code into a wide range of software projects, including proprietary software. We present here a montaging software package \u2026\n\n\n---\n\n## 3. Framework for the Generation of 3D Fiber Composite Structures from 2D Observations\nValid Text Names:\n'clarke2024frameworkforthe pages 1-2', 'clarke2024frameworkforthe pages 2-4', 'clarke2024frameworkforthe pages 4-6', 'clarke2024frameworkforthe pages 6-7', 'clarke2024frameworkforthe pages 7-9', 'clarke2024frameworkforthe pages 9-13', 'clarke2024frameworkforthe pages 13-15', 'clarke2024frameworkforthe pages 15-16', 'clarke2024frameworkforthe pages 16-16'\n\nBibTex:\n@article{clarke2024frameworkforthe,\n author = \"Clarke, Kenneth M. and Wertz, John and Groeber, Michael\",\n title = \"Framework for the Generation of 3D Fiber Composite Structures from 2D Observations\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"13\",\n pages = \"569-584\",\n month = \"May\",\n doi = \"10.1007/s40192-024-00352-8\",\n url = \"https://doi.org/10.1007/s40192-024-00352-8\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nSynthetic generation of realistic materials for testing of process\u2013structure\u2013property (PSP) relationships in computational materials science has gained significant traction over the past two decades. Generation tools continue to lag in some aspects of realism, leading to uncertainty and errors in simulating material response. The experimental collection of information to guide generation of 3D synthetic structures remains costly, time consuming, and challenging. These challenges are compounded by limitations of stereology, which permits estimation of 3D microstructural statistics from 2D observations under restrictive assumptions on constituent morphologies and size distributions and can be difficult to apply to microstructure metrics like constituent orientation distribution and clustering. This work seeks to overcome these challenges by introducing a framework for learning probable 3D microstructure statistics by minimizing the loss determined by matching statistics obtained from 2D observations via synthetic microstructure generation software (e.g. Dream.3D) and stereological principles. This framework is applied to short carbon fiber composite structures printed from an additive manufacturing process, Direct Ink Writing, with the hope that the framework could generalize to other particulate structures as well as allow for simulation and design of materials structure.\n\nRelevant Snippet:\n\u2026 Structures from 2D Observations | Integrating Materials and Manufacturing Innovation \u2026 from 2D observations via synthetic microstructure generation software (eg Dream.3D) and \u2026\n\n\n---\n\n## 4. AFRL Additive Manufacturing Modeling Series: Challenge 4, 3D Reconstruction of an IN625 High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning\nValid Text Names:\n'chapman2021afrladditivemanufacturing pages 1-3', 'chapman2021afrladditivemanufacturing pages 3-5', 'chapman2021afrladditivemanufacturing pages 5-7', 'chapman2021afrladditivemanufacturing pages 7-8', 'chapman2021afrladditivemanufacturing pages 8-10', 'chapman2021afrladditivemanufacturing pages 10-11', 'chapman2021afrladditivemanufacturing pages 11-13', 'chapman2021afrladditivemanufacturing pages 13-13'\n\nBibTex:\n@article{chapman2021afrladditivemanufacturing,\n author = \"Chapman, Michael G. and Shah, Megna N. and Donegan, Sean P. and Scott, J. Michael and Shade, Paul A. and Menasche, David and Uchic, Michael D.\",\n title = \"AFRL Additive Manufacturing Modeling Series: Challenge 4, 3D Reconstruction of an IN625 High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning\",\n year = \"2021\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"10\",\n pages = \"129-141\",\n month = \"Jun\",\n doi = \"10.1007/s40192-021-00212-9\",\n url = \"https://doi.org/10.1007/s40192-021-00212-9\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nHigh-energy diffraction microscopy (HEDM) in-situ mechanical testing experiments offer unique insight into the evolving deformation state within polycrystalline materials. These experiments rely on a sophisticated analysis of the diffraction data to instantiate a 3D reconstruction of grains and other microstructural features associated with the test volume. For microstructures of engineering alloys that are highly twinned and contain numerous features around the estimated spatial resolution of HEDM reconstructions, the accuracy of the reconstructed microstructure is not known. In this study, we address this uncertainty by characterizing the same HEDM sample volume using destructive serial sectioning (SS) that has higher spatial resolution. The SS experiment was performed on an Inconel 625 alloy sample that had undergone HEDM in-situ mechanical testing to a small amount of plastic strain (~\u20090.7%), which was part of the Air Force Research Laboratory Additive Manufacturing (AM) Modeling Series. A custom-built automated multi-modal SS system was used to characterize the entire test volume, with a spatial resolution of approximately 1 \u00b5m. Epi-illumination optical microscopy images, backscattered electron images, and electron backscattered diffraction maps were collected on every section. All three data modes were utilized and custom data fusion protocols were developed for 3D reconstruction of the test volume. The grain data were homogenized and downsampled to 2 \u00b5m as input for Challenge 4 of the AM Modeling Series, which is available at the Materials Data Facility repository.\n\nRelevant Snippet:\n\u2026 3D file, and can be viewed with any software program that is capable of reading and viewing an HDF5 file. In addition, if the text in the \u2018Pipeline\u2019 group is saved as a new *.json file, this \u2026\n\n\n---\n\n## 5. AixViPMaP\u00ae\u2014an Operational Platform for Microstructure Modeling Workflows\nValid Text Names:\n'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 1-2', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 2-3', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 3-5', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 5-8', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 8-11', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 11-12', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 12-14', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 14-17', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 17-19', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 19-21', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 21-22', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 22-22'\n\nBibTex:\n@article{koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform,\n author = \"Koschmieder, L. and Hojda, S. and Apel, M. and Altenfeld, R. and Bami, Y. and Haase, C. and Lin, M. and Vuppala, A. and Hirt, G. and Schmitz, G.J.\",\n title = \"AixViPMaP\u00ae\u2014an Operational Platform for Microstructure Modeling Workflows\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"122-143\",\n month = \"Jun\",\n doi = \"10.1007/s40192-019-00138-3\",\n url = \"https://doi.org/10.1007/s40192-019-00138-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 supporting some but not all simulation software. For a co-simulation framework to support software, usually, the data model of the specific software must be specified and additional APIs \u2026\n\n\nUnobtainable Papers:\nR Brian Leavy, John Clayton, Jonathan Ligda, and Jaroslaw Knap. Methods for microstructure modeling and phase field simulations of fracture. ArXiv, Jun 2023. URL: https://doi.org/10.21236/ad1203858, doi:10.21236/ad1203858.\nG. J. Schmitz. Microstructure modeling in integrated computational materials engineering (icme) settings: can hdf5 provide the basis for an emerging standard for describing microstructures? JOM, 68:77-83, Dec 2016. URL: https://doi.org/10.1007/s11837-015-1748-2, doi:10.1007/s11837-015-1748-2.\nStanislav Buklovskyi, Kateryna Miroshnichenko, Igor Tsukrov, Rebecca J. Thomson, Peder C. Solberg, and Douglas W. Van Citters. Effective thermal conductivity of carbon/ultra-high-molecular-weight-polyethylene nanocomposites. Procedia Structural Integrity, 80:146-156, Jan 2026. URL: https://doi.org/10.1016/j.prostr.2026.02.014, doi:10.1016/j.prostr.2026.02.014.\nCarter K. Cocke, Anthony D. Rollett, Ricardo A. Lebensohn, and Ashley D. Spear. The afrl additive manufacturing modeling challenge: predicting micromechanical fields in am in625 using an fft-based method with direct input from a 3d microstructural image. Integrating Materials and Manufacturing Innovation, 10:157-176, May 2021. URL: https://doi.org/10.1007/s40192-021-00211-w, doi:10.1007/s40192-021-00211-w.\nAlexander Ra\u00dfloff, Paul Schulz, Robert K\u00fchne, Marreddy Ambati, Ilja Koch, Andr\u00e9 T. Zeuner, Maik Gude, Martina Zimmermann, and Markus K\u00e4stner. Experimental-numerical analysis of microstructure-property linkages for additively manufactured materials. Procedia Structural Integrity, 38:4-11, Jan 2022. URL: https://doi.org/10.1016/j.prostr.2022.03.002, doi:10.1016/j.prostr.2022.03.002.\nM. Jackson, M. Groeber, S. Donegan, and D. Dimiduk. Advancements to the digital representation environment for analysis of materials in 3-dimensions\u2014dream.3d. Microscopy and Microanalysis, 25:176-177, Aug 2019. URL: https://doi.org/10.1017/s1431927619001612, doi:10.1017/s1431927619001612.\nJonathan Robichaud, Tim Vincent, Ben Schultheis, and Anil Chaudhary. Integrated computational materials engineering to predict melt-pool dimensions and 3d grain structures for selective laser melting of inconel 625. Integrating Materials and Manufacturing Innovation, 8:305-317, Jul 2019. URL: https://doi.org/10.1007/s40192-019-00145-4, doi:10.1007/s40192-019-00145-4.\nAyman A. Salem, Joshua B. Shaffer, Richard A. Kublik, Luke A. Wuertemberger, and Daniel P. Satko. Microstructure-informed cloud computing for interoperability of materials databases and computational models: microtextured regions in ti alloys. Mar 2017. URL: https://doi.org/10.1007/s40192-017-0090-7, doi:10.1007/s40192-017-0090-7.\nJohn Wertz, Erik P. Blasch, Laura Homa, and Matt Cherry. Dynamic Data Driven Applications Systems Analysis of Microtexture Regions in Titanium Alloys, pages 199-222. Springer Nature Switzerland, Jan 2026. URL: https://doi.org/10.1007/978-3-031-88574-7\\_7, doi:10.1007/978-3-031-88574-7\\_7.\nBin Li, Tao Gao, Zhi Qin, Hongqian Xue, and Zhidan Sun. Investigation of the fatigue life scatter for aa7075-t6 using crystal plasticity finite element method in the high to very high cycle fatigue regime. Integrating Materials and Manufacturing Innovation, 11:198-213, Mar 2022. URL: https://doi.org/10.1007/s40192-022-00256-5, doi:10.1007/s40192-022-00256-5.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_Z5I4Dszqv8IduAWdkGxzojFr", + "info": { + "start_ts": 703.492846157, + "end_ts": 739.275643191 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\nValid Text Names:\n'savage2021anautomatedprocedure pages 1-2', 'savage2021anautomatedprocedure pages 2-3', 'savage2021anautomatedprocedure pages 3-4', 'savage2021anautomatedprocedure pages 4-5', 'savage2021anautomatedprocedure pages 5-6', 'savage2021anautomatedprocedure pages 6-7', 'savage2021anautomatedprocedure pages 7-9', 'savage2021anautomatedprocedure pages 9-11', 'savage2021anautomatedprocedure pages 11-13', 'savage2021anautomatedprocedure pages 13-14', 'savage2021anautomatedprocedure pages 14-15', 'savage2021anautomatedprocedure pages 15-16', 'savage2021anautomatedprocedure pages 16-16'\n\nBibTex:\n@article{savage2021anautomatedprocedure,\n author = \"Savage, Daniel J. and McCabe, Rodney J. and Knezevic, Marko\",\n title = \"An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\",\n volume = \"171\",\n issn = \"1044-5803\",\n url = \"https://doi.org/10.1016/j.matchar.2020.110808\",\n doi = \"10.1016/j.matchar.2020.110808\",\n journal = \"Materials Characterization\",\n publisher = \"Elsevier BV\",\n year = \"2021\",\n month = \"Jan\",\n pages = \"110808\"\n}\n\n\nRelevant Snippet:\n\u2026 or neutron diffraction bulk texture measurements based on the quantitative change in fraction of c-axes within some angle to the compression direction [17, 53]. The texture method is \u2026\n\n\n---\n\n## 2. A Survey and Comparison of Methods for Quantitative TextureAnalysis\nValid Text Names:\n'einhorn2017asurveyand pages 1-5', 'einhorn2017asurveyand pages 5-7', 'einhorn2017asurveyand pages 7-11', 'einhorn2017asurveyand pages 11-16', 'einhorn2017asurveyand pages 16-20', 'einhorn2017asurveyand pages 20-23', 'einhorn2017asurveyand pages 23-27', 'einhorn2017asurveyand pages 27-30', 'einhorn2017asurveyand pages 30-35', 'einhorn2017asurveyand pages 35-41', 'einhorn2017asurveyand pages 41-46', 'einhorn2017asurveyand pages 46-52', 'einhorn2017asurveyand pages 52-56', 'einhorn2017asurveyand pages 56-60', 'einhorn2017asurveyand pages 60-63', 'einhorn2017asurveyand pages 67-71', 'einhorn2017asurveyand pages 71-75', 'einhorn2017asurveyand pages 79-84', 'einhorn2017asurveyand pages 88-92', 'einhorn2017asurveyand pages 92-95', 'einhorn2017asurveyand pages 95-98', 'einhorn2017asurveyand pages 98-102', 'einhorn2017asurveyand pages 102-104', 'einhorn2017asurveyand pages 104-107', 'einhorn2017asurveyand pages 109-109'\n\nBibTex:\n@article{einhorn2017asurveyand,\n author = \"Einhorn, John\",\n title = \"A Survey and Comparison of Methods for Quantitative TextureAnalysis\",\n year = \"2017\",\n journal = \"ArXiv\",\n doi = \"10.18130/v30r4d\",\n url = \"https://doi.org/10.18130/v30r4d\",\n month = \"Jul\",\n publisher = \"University of Virginia\"\n}\n\n\nAbstract:\nNumerous methods for measuring and analyzing texture\nare explored and discussed. Traditional pole figure inversion\nmethods are compared against more recent developments in\nquantitative texture analysis involving the Rietveld whole pattern\nrefinement method. A modified equal area pole figure collection\nstrategy is compared to the conventional equiangular method using\nlaboratory X-ray diffraction (XRD). The two measurement strategies\nwere found to yield qualitatively similar orientation distribution\nfunctions (ODF), and the former set takes about half the time to\ncollect. However, the estimated texture was found to be strongly\ndependent on the ODF computation parameters used in MTEX. The\ntexture evolution during compression of a high specific strength\nMg-base alloy containing long period stacking order (LPSO) phase is\nstudied using synchrotron XRD. The initial texture was found to be\nmuch sharper than previously published results, demonstrating the\nimportance of using a \u201cdirect method\u201d to compute the OD for this\nmaterial. The evolution of the texture of the Mg and LPSO phases\nsuggests that the deformation is accommodated almost entirely by\ntwinning in the former case and by dislocation slip, elastic\ndeformation, and possibly some twinning in the latter. The texture\nof \u03b1-uranium foils is measured via X-ray (both Cu- and Mo-K\u03b1\nradiation) and neutron diffraction, and the results are analyzed\nwith MTEX and MAUD (Rietveld), respectively. The three radiation\nsources are found to yield slightly different results, based on\nprobing different depths of the foils. Finally, the capability of\nmeasuring texture on the Neutron Residual Stress Mapping Facility\n(NRSF2) and Nanoscale Ordered Materials Diffractometer (NOMAD)\ninstruments at the Oak Ridge National Laboratory (ORNL) is\ndemonstrated by comparing measurements to those obtained via\nconventional XRD methods. Pole figure inversion in MTEX (for XRD\nand NRSF2) and Rietveld analysis in MAUD (for NOMAD) are employed\nfor analysis, and the results are quantitatively compared.\nReasonable agreement is found across all methods, and the\ndifferences between methods are attributed mostly to differences in\nthe analysis procedure. The implications of slight variations in\ntexture, as estimated by the various techniques, are explored\nthrough calculations of tensor properties and comparisons with\npredicted deformation texture evolutions. Finally, suggestions for\nimproving the data collection speed by collecting less data are\nexplored for both instruments.\n\nRelevant Snippet:\n\u2026 However, the estimated texture was found to be strongly dependent on the ODF computation parameters used in MTEX. The texture evolution during compression of a high specific \u2026\n\n\n---\n\n## 3. Quantitative texture analysis and combined analysis\nValid Text Names:\n'chateigner2019quantitativetextureanalysis pages 1-2', 'chateigner2019quantitativetextureanalysis pages 2-3', 'chateigner2019quantitativetextureanalysis pages 3-4', 'chateigner2019quantitativetextureanalysis pages 4-5', 'chateigner2019quantitativetextureanalysis pages 5-6', 'chateigner2019quantitativetextureanalysis pages 6-7', 'chateigner2019quantitativetextureanalysis pages 7-8', 'chateigner2019quantitativetextureanalysis pages 8-10', 'chateigner2019quantitativetextureanalysis pages 10-11', 'chateigner2019quantitativetextureanalysis pages 11-12', 'chateigner2019quantitativetextureanalysis pages 12-13', 'chateigner2019quantitativetextureanalysis pages 13-15', 'chateigner2019quantitativetextureanalysis pages 15-16', 'chateigner2019quantitativetextureanalysis pages 16-17', 'chateigner2019quantitativetextureanalysis pages 17-18', 'chateigner2019quantitativetextureanalysis pages 18-19', 'chateigner2019quantitativetextureanalysis pages 18-18', 'chateigner2019quantitativetextureanalysis pages 19-21', 'chateigner2019quantitativetextureanalysis pages 21-22', 'chateigner2019quantitativetextureanalysis pages 22-23', 'chateigner2019quantitativetextureanalysis pages 23-24', 'chateigner2019quantitativetextureanalysis pages 24-25', 'chateigner2019quantitativetextureanalysis pages 25-26', 'chateigner2019quantitativetextureanalysis pages 26-26'\n\nBibTex:\n@inbook{chateigner2019quantitativetextureanalysis,\n author = \"Chateigner, D. and Lutterotti, L. and Morales, M.\",\n title = \"Quantitative texture analysis and combined analysis\",\n ISBN = \"9781118416280\",\n url = \"https://doi.org/10.1107/97809553602060000968\",\n doi = \"10.1107/97809553602060000968\",\n booktitle = \"International Tables for Crystallography\",\n publisher = \"International Union of Crystallography\",\n year = \"2019\",\n month = \"Dec\",\n pages = \"555-580\",\n journal = \"International Tables for Crystallography\"\n}\n\n\nRelevant Snippet:\n\u2026 Here, we illustrate magnetic quantitative texture analysis on an iron sample exhibiting magnetic saturation around 0.5T (Chateigner et al., 2010) that was initially not magnetized. \u2026\n\n\nUnobtainable Papers:\nS. Gorelova, H. Schaeben, and R. Kawalla. Quantifying texture evolution during hot rolling of magnesium twin roll cast strip. Materials Science and Engineering A-structural Materials Properties Microstructure and Processing, 602:134-142, Apr 2014. URL: https://doi.org/10.1016/j.msea.2014.02.057, doi:10.1016/j.msea.2014.02.057.\nFlorencia Malamud, Steve Gaudez, Jan Capek, Ezequiel Fogliatto, and Markus Strobl. Texture analysis implementation at the neutron strain diffractometer poldi. Materials Characterization, Apr 2025. URL: https://doi.org/10.1016/j.matchar.2025.115003, doi:10.1016/j.matchar.2025.115003.\nS. Gorelova and H. Schaeben. Quantifying texture evolution during hot rolling of az31 twin roll cast strip. IOP Conference Series: Materials Science and Engineering, 82:012003, Apr 2015. URL: https://doi.org/10.1088/1757-899x/82/1/012003, doi:10.1088/1757-899x/82/1/012003.\nGennady Rafailov, El'ad N. Caspi, Ralf Hielscher, Eitan Tiferet, Roni Schneck, and Sven C. Vogel. Visualization of texture components using mtex. Journal of Applied Crystallography, 53:540-548, Feb 2020. URL: https://doi.org/10.1107/s1600576719014742, doi:10.1107/s1600576719014742.\nRoberto Visalli, Gaetano Ortolano, Gaston Godard, and Rosolino Cirrincione. Micro-fabric analyzer (mfa): a new semiautomated arcgis-based edge detector for quantitative microstructural analysis of rock thin-sections. ISPRS Int. J. Geo Inf., 10:51, Jan 2021. URL: https://doi.org/10.3390/ijgi10020051, doi:10.3390/ijgi10020051.\nDavid Mainprice, Ralf Hielscher, and Helmut Schaeben. Calculating anisotropic physical properties from texture data using the mtex open-source package. Geological Society, London, Special Publications, 360:175-192, Jan 2011. URL: https://doi.org/10.1144/sp360.10, doi:10.1144/sp360.10.\nFlorian Bachmann, Ralf Hielscher, and Helmut Schaeben. Grain detection from 2d and 3d ebsd data\u2013specification of the mtex algorithm. Ultramicroscopy, 111 12:1720-33, Dec 2011. URL: https://doi.org/10.1016/j.ultramic.2011.08.002, doi:10.1016/j.ultramic.2011.08.002.\nM. Y. Xie, N. Baimpas, C. Reinhard, and A. M. Korsunsky. Texture analysis in cubic phase polycrystals by single exposure synchrotron x-ray diffraction. Journal of Applied Physics, 114:163502, Oct 2013. URL: https://doi.org/10.1063/1.4825120, doi:10.1063/1.4825120.\nV. A. Yardley, S. Fahimi, and E. J. Payton. Classification of creep crack and cavitation sites in tempered martensite ferritic steel microstructures using mtex toolbox for ebsd. Materials Science and Technology, 31:547-553, Mar 2015. URL: https://doi.org/10.1179/1743284714y.0000000603, doi:10.1179/1743284714y.0000000603.\nDavid Mainprice, Florian Bachmann, Ralf Hielscher, and Helmut Schaeben. Descriptive tools for the analysis of texture projects with large datasets using mtex: strength, symmetry and components. Geological Society, London, Special Publications, 409:251-271, Aug 2015. URL: https://doi.org/10.1144/sp409.8, doi:10.1144/sp409.8.\nDavid Mainprice, Florian Bachmann, Ralf Hielscher, Helmut Schaeben, and Geoffrey E. Lloyd. Calculating anisotropic piezoelectric properties from texture data using the mtex open source package. Geological Society, London, Special Publications, 409:223-249, Jul 2015. URL: https://doi.org/10.1144/sp409.2, doi:10.1144/sp409.2.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_5hiGgVfNiBeJ5xEkpggCPT0O", + "info": { + "start_ts": 703.531510634, + "end_ts": 741.917590527 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nAndrea M. Jokisaari, Khadija Mahbuba, Yuhao Wang, and Benjamin Beeler. The impact of anisotropic thermal expansion on the isothermal annealing of polycrystalline \u03b1-uranium. Computational Materials Science, 205:111217, Apr 2022. URL: https://doi.org/10.1016/j.commatsci.2022.111217, doi:10.1016/j.commatsci.2022.111217.\nHenry Proudhon, Nicolas Gu\u00e9ninchault, Samuel Forest, and Wolfgang Ludwig. Incipient bulk polycrystal plasticity observed by synchrotron in-situ topotomography. Materials, 11:2018, Oct 2018. URL: https://doi.org/10.3390/ma11102018, doi:10.3390/ma11102018.\nJuan Manuel Garc\u00eda and Thilo F. Morgeneyer. Strength and fatigue strength of a similar ti\u20106al\u20102sn\u20104zr\u20102mo\u20100.1si linear friction welded joint. Fatigue & Fracture of Engineering Materials & Structures, 42:1100-1117, Mar 2019. URL: https://doi.org/10.1111/ffe.12973, doi:10.1111/ffe.12973.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_t3X9GkiKh5YtIwvvFGFtqFju", + "info": { + "start_ts": 703.565379123, + "end_ts": 745.716529263 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. A Roadmap for Edge Computing Enabled Automated Multidimensional Transmission Electron Microscopy\nValid Text Names:\n'mukherjee2022aroadmapfor pages 1-2', 'mukherjee2022aroadmapfor pages 2-3', 'mukherjee2022aroadmapfor pages 3-4', 'mukherjee2022aroadmapfor pages 4-6', 'mukherjee2022aroadmapfor pages 6-8', 'mukherjee2022aroadmapfor pages 8-9', 'mukherjee2022aroadmapfor pages 9-10', 'mukherjee2022aroadmapfor pages 10-11', 'mukherjee2022aroadmapfor pages 11-11'\n\nBibTex:\n@article{mukherjee2022aroadmapfor,\n author = \"Mukherjee, Debangshu and Roccapriore, Kevin M. and Al-Najjar, Anees and Ghosh, Ayana and Hinkle, Jacob D. and Lupini, Andrew R. and Vasudevan, Rama K. and Kalinin, Sergei V. and Ovchinnikova, Olga S. and Ziatdinov, Maxim A. and Rao, Nageswara S.\",\n title = \"A Roadmap for Edge Computing Enabled Automated Multidimensional Transmission Electron Microscopy\",\n year = \"2022\",\n journal = \"Microscopy Today\",\n volume = \"30\",\n pages = \"10-19\",\n month = \"Oct\",\n doi = \"10.48550/arxiv.2210.02538\",\n url = \"https://doi.org/10.48550/arxiv.2210.02538\",\n publisher = \"arXiv\"\n}\n\n\nAbstract:\nAbstract: The advent of modern, high-speed electron detectors has made the collection of multidimensional hyperspectral transmission electron microscopy datasets, such as 4D-STEM, a routine. However, many microscopists find such experiments daunting since analysis, collection, long-term storage, and networking of such datasets remain challenging. Some common issues are their large and unwieldy size that often are several gigabytes, non-standardized data analysis routines, and a lack of clarity about the computing and network resources needed to utilize the electron microscope. The existing computing and networking bottlenecks introduce significant penalties in each step of these experiments, and thus, real-time analysis-driven automated experimentation for multidimensional TEM is challenging. One solution is to integrate microscopy with edge computing, where moderately powerful computational hardware performs the preliminary analysis before handing off the heavier computation to high-performance computing (HPC) systems. Here we trace the roots of computation in modern electron microscopy, demonstrate deep learning experiments running on an edge system, and discuss the networking requirements for tying together microscopes, edge computers, and HPC systems.\n\nRelevant Snippet:\n\u2026 Elastically scattered electron diffraction patterns (the 4D-\u2026 allows flexible Python-based control of the microscope hardware \u2026 This has led to development of open-source microscope data \u2026\n\n\n---\n\n## 2. Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook.\nValid Text Names:\n'botifoll2022machinelearningin pages 1-2', 'botifoll2022machinelearningin pages 3-4', 'botifoll2022machinelearningin pages 4-6', 'botifoll2022machinelearningin pages 6-7', 'botifoll2022machinelearningin pages 7-8', 'botifoll2022machinelearningin pages 8-9', 'botifoll2022machinelearningin pages 9-10', 'botifoll2022machinelearningin pages 10-12', 'botifoll2022machinelearningin pages 13-14', 'botifoll2022machinelearningin pages 14-15', 'botifoll2022machinelearningin pages 17-18', 'botifoll2022machinelearningin pages 19-21', 'botifoll2022machinelearningin pages 21-23', 'botifoll2022machinelearningin pages 23-24', 'botifoll2022machinelearningin pages 26-27', 'botifoll2022machinelearningin pages 30-31', 'botifoll2022machinelearningin pages 31-32', 'botifoll2022machinelearningin pages 32-33', 'botifoll2022machinelearningin pages 33-35', 'botifoll2022machinelearningin pages 35-36', 'botifoll2022machinelearningin pages 40-41', 'botifoll2022machinelearningin pages 42-43', 'botifoll2022machinelearningin pages 45-46', 'botifoll2022machinelearningin pages 47-48', 'botifoll2022machinelearningin pages 50-51'\n\nBibTex:\n@article{botifoll2022machinelearningin,\n author = \"Botifoll, Marc and Pinto-Huguet, Ivan and Arbiol, J.\",\n title = \"Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook.\",\n year = \"2022\",\n journal = \"Nanoscale horizons\",\n month = \"Oct\",\n doi = \"10.1039/d2nh00377e\",\n url = \"https://doi.org/10.1039/d2nh00377e\"\n}\n\n\nAbstract:\nIn the last few years, electron microscopy has experienced a new methodological paradigm aimed to fix the bottlenecks and overcome the challenges of its analytical workflow. Machine learning and artificial intelligence are answering this call providing powerful resources towards automation, exploration, and development. In this review, we evaluate the state-of-the-art of machine learning applied to electron microscopy (and obliquely, to materials and nano-sciences). We start from the traditional imaging techniques to reach the newest higher-dimensionality ones, also covering the recent advances in spectroscopy and tomography. Additionally, the present review provides a practical guide for microscopists, and in general for material scientists, but not necessarily advanced machine learning practitioners, to straightforwardly apply the offered set of tools to their own research. To conclude, we explore the state-of-the-art of other disciplines with a broader experience in applying artificial intelligence methods to their research (e.g., high-energy physics, astronomy, Earth sciences, and even robotics, videogames, or marketing and finances), in order to narrow down the incoming future of electron microscopy, its challenges and outlook.\n\n\nUnobtainable Papers:\nGear\u00f3id Liam Mangan, Grigore Moldovan, and Andrew Stewart. Influence: an open-source python package to model images captured with direct electron detectors. Microscopy and microanalysis : the official journal of Microscopy Society of America, Microbeam Analysis Society, Microscopical Society of Canada, 29 4:1380-1401, Jul 2023. URL: https://doi.org/10.1093/micmic/ozad064, doi:10.1093/micmic/ozad064.\nBenjamin Miller, Anahita Pakzad, Benjamin Savitzky, Colin Ophus, and Cory Czarnik. Using py4dstem in gms: hybrid open-source, commercial-freeware methods for analyzing 4d stem datasets. Microscopy and Microanalysis, 27:1352-1354, Jul 2021. URL: https://doi.org/10.1017/s1431927621005043, doi:10.1017/s1431927621005043.\nRandi Holmestad, Elisabeth Thronsen, Yasuhito Kawahara, Tina Bergh, J\u00f8rgen A S\u00f8rhaug, Christoph M Hell, Ruben Bj\u00f8rge, Emil F Christiansen, Kenji Kaneko, and Calin D Marioara. Nucleation and phase development of precipitates in age-hardenable aluminium alloys studied by 4d-stem. Microscopy and microanalysis : the official journal of Microscopy Society of America, Microbeam Analysis Society, Microscopical Society of Canada, 29 Supplement_1:305-306, Jul 2023. URL: https://doi.org/10.1093/micmic/ozad067.141, doi:10.1093/micmic/ozad067.141.\nH. W. Leung, R. C. B. Copley, J. Laulainen, D. N. Johnstone, and P. A. Midgley. Combining scanning nanobeam electron diffraction with 3d electron diffraction to investigate crystal defects. Microscopy and Microanalysis, Jul 2024. URL: https://doi.org/10.1093/mam/ozae044.901, doi:10.1093/mam/ozae044.901.\nFernando C Castro and Anahita Pakzad. Optimizing 4d stem data collection and processing for crystal orientation mapping analysis in digitalmicrograph. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.096, doi:10.1093/mam/ozaf048.096.\nGear\u00f3id Liam Mangan. Advancements and applications of (scanning) transmission electron microscopy simulations. Text, Jan 2022. URL: https://doi.org/10.34961/researchrepository-ul.21621174, doi:10.34961/researchrepository-ul.21621174.\nPavlina Sikorova, Miroslav Slouf, Tomas Molnar, and Vladislav Krzyzanek. Computational methods for automated center determination in electron diffraction patterns. Journal of Applied Crystallography, 59(3):845-857, Apr 2026. URL: https://doi.org/10.1107/s1600576726002384, doi:10.1107/s1600576726002384.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_2WFpaGkkorc7QNUjgoGtFJOi", + "info": { + "start_ts": 703.598195246, + "end_ts": 765.667408384 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nDeveloping a novel hierarchical approach for multiscale structural reliability predictions for ultra-high consequence applications\nINVESTIGATION OF SINGLE CRYSTAL AND BI-CRYSTAL DEFORMATION IN BODY-CENTERED CUBIC TANTALUM USING INDENTATION\nCaleb E. Andrews, Maria Strantza, Nicholas P. Calta, Manyalibo J. Matthews, and Mitra L. Taheri. A denoising autoencoder for improved kikuchi pattern quality and indexing in electron backscatter diffraction. Ultramicroscopy, 253:113810, Nov 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113810, doi:10.1016/j.ultramic.2023.113810.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_xqX7V5hmrKT4tjzjK7lLSxlE", + "info": { + "start_ts": 703.631840635, + "end_ts": 739.262894677 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_QhQG5TaOdTjPfKLBDrG4HStR", + "type": "function", + "function": { + "arguments": "{\"question\": \"Extract evidence relevant to comparing EBSD post-processing tools: (1) MTEX grain reconstruction and texture/ODF capabilities; (2) DREAM.3D grain segmentation and 2D/3D EBSD processing; (3) use of KAM or grain boundaries in EBSD workflows; (4) peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, PyMicro. Include exact capabilities and limitations where stated.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Most relevant evidence (with ids):\n\n## pqac-00000000\n\nFrom text 'folwarczny2026accurategrainboundary pages 5-9'.\n\nMartin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal.\n\nThe excerpt describes using MTEX and DREAM.3D for EBSD analysis. MTEX capabilities include: plotting inverse pole figure (IPF) maps and orientation distribution function (ODF) pole figures for texture analysis, grain size distribution analysis using equivalent circle radius function with lognormal fitting, grain shape analysis via ellipse fitting with aspect ratio and characteristic shape function calculations, and grain boundary segment orientation analysis. DREAM.3D is used for simulating 3D microstructures and exporting 2D planar sections in multiple orthogonal directions (XY, XZ, YZ) with proper reference frame rotations to match experimental EBSD conventions. The workflow converts DREAM.3D output to standard .ang format used in TSL OIM Analysis. No information is provided regarding KAM usage, orix, DefDAP, kikuchipy, pyxem, or PyMicro tools.\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000]\n\nStatus: Paper Count=22 | Relevant Papers=1 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=1 | Current Cost=$0.0899 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_QhQG5TaOdTjPfKLBDrG4HStR", + "info": { + "start_ts": 779.597149611, + "end_ts": 787.125253756 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_at4kevxNAZzrz4Vmrs3X4MTb", + "type": "function", + "function": { + "arguments": "{\"question\": \"What does the literature say specifically about DREAM.3D grain segmentation algorithms, misorientation thresholds, 2D/3D EBSD reconstruction, automation/pipelines, and maturity?\"}", + "name": "gather_evidence" + } + }, + { + "id": "call_1oiOo0bOuigk4ZUPXbdj7xZ2", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"polonsky2019threedimensionalanalysisand pages 1-2\", \"polonsky2019threedimensionalanalysisand pages 2-5\", \"stanek2020comparisonofsegmentation pages 1-2\", \"stanek2020comparisonofsegmentation pages 2-7\", \"savage2021anautomatedprocedure pages 1-2\", \"wanni2024machinelearningenhanced pages 1-2\"]}", + "name": "read_text" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Most relevant evidence (with ids):\n\n## pqac-00000001\n\nFrom text 'polonsky2019threedimensionalanalysisand pages 1-2'.\n\nAndrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\nThe excerpt discusses DREAM.3D as a mature software package for 3D EBSD data reconstruction and microstructure analysis that aggregates community-developed algorithms. It operates as a multi-step pipeline with numerous input variables and outputs. Traditionally, DREAM.3D runs serially on workstations with predefined parameters. The authors integrated DREAM.3D into a cloud-based platform (BisQue) to enable parallelization and parameterization of pipeline variables across value ranges. This parallelization specifically provided insights about grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets of additively manufactured materials. The manual, serial approach previously limited exploration of the full parameter space due to long computation times scaling with dataset size and complexity, introducing bias in reconstruction processes.\n\n\n---\n\n## pqac-00000002\n\nFrom text 'polonsky2019threedimensionalanalysisand pages 13-14'.\n\nAndrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\nThe excerpt discusses DREAM.3D's integration into the BisQue cloud platform for 3D microstructure reconstruction. Key findings include: (1) grain segmentation misorientation tolerance optimization specifically for additively manufactured materials yielded new insights; (2) algorithms for 2D and 3D volume element sampling revealed that traditional 2D metallographic sectioning produces grain size measurements approximately half those of equivalent 3D measurements in heterogeneous AM microstructures; (3) the platform enables pipeline parameterization and parallelized module execution for reconstruction optimization; (4) automation includes massively parallelized slicing/cropping pipelines to simulate 2D versus 3D statistical grain measurements. The work demonstrates DREAM.3D's practical application for addressing discrepancies between 2D and 3D grain analysis in additively manufactured materials.\n\n\n---\n\n## pqac-00000003\n\nFrom text 'polonsky2019threedimensionalanalysisand pages 2-5'.\n\nAndrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\nThe excerpt describes DREAM.3D's application for 3D EBSD reconstruction and grain segmentation in additively manufactured materials. Grains are segmented using misorientation tolerance thresholds, with specific implementation details provided for 304L stainless steel (misorientation tolerance with minimum grain size of 64 voxels) and another alloy (2\u00b0 misorientation tolerance with 125 voxels minimum). The tool supports 2D EBSD slice reconstruction into coherent 3D volumes. Key automation features include parameterizable pipeline execution allowing parallel processing across multiple compute nodes with variable parameter ranges rather than discrete values. This parallelization capability reduces setup and analysis time for parameter studies. The excerpt indicates DREAM.3D was integrated with BisQue cloud infrastructure, suggesting mature implementation for workflow automation and data provenance tracking, though the excerpt does not extensively discuss algorithm maturity or theoretical foundations.\n\n\n---\n\n## pqac-00000004\n\nFrom text 'yi2025rapidgrainsegmentation pages 4-7'.\n\nYu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal.\n\nThe excerpt describes Dream3D's application to 2D/3D EBSD reconstruction from the Small IN100 dataset (117 slices of EBSD data with 189 x 201 scan points). Dream3D employs automated pipelines for grain boundary reconstruction, including full reconstruction, morphological statistics, crystallographic statistics, mesh generation, and grain boundary character distribution (GBCD) analysis. The segmentation achieved high performance metrics (average IoU of 91.58% and Dice score of 95.6%). However, limitations exist: the algorithm relies on Inverse Pole Figure (IPF) color contrast and can misidentify neighboring grains with similar IPF colors as the same grain, causing missegmentation. The excerpt acknowledges this as an inherent uncertainty when using this pipeline for grain segmentation. No specific information about misorientation thresholds is provided. The text notes that EBSD remains necessary when precise grain orientation information is required, suggesting Dream3D is best suited for rapidly quantifying grain size distribution with approximately 10% error tolerance.\n\n\n---\n\n## pqac-00000005\n\nFrom text 'stanek2020comparisonofsegmentation pages 1-2'.\n\nJ. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\nThe excerpt mentions DREAM.3D software as a computational tool that enables significant data evaluation and five-parameter grain boundary analysis, though it does not elaborate on specific segmentation algorithms. Regarding misorientation thresholds, the paper notes that typical values are 5\u00b0 in EDAX OIM software, with other choices possible. The work demonstrates critical differences between 2D and 3D EBSD segmentation: identical misorientation thresholds produce different results in 2D versus 3D cases because nonconvex grains may appear as separate grains in 2D slices but belong to single grains in 3D reconstruction. The paper emphasizes that proper segmentation attention is necessary to avoid misleading quantitative microstructure results. However, the excerpt provides limited information about automation/pipelines or the overall maturity level of these technologies.\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005]\n\nStatus: Paper Count=22 | Relevant Papers=4 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=6 | Current Cost=$0.1324 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_at4kevxNAZzrz4Vmrs3X4MTb", + "info": { + "start_ts": 805.10017757, + "end_ts": 814.30554037 + } + }, + { + "role": "tool", + "content": "## 1. wanni2024machinelearningenhanced pages 1-2 (Context ID: pqac-00000006)\n\n npj | computational materials Article\n\n Published in partnership with the Shanghai Institute of Ceramics of the Chinese Academy of Sciences\n\n\n\n\n\n https://doi.org/10.1038/s41524-024-01324-4\n Machine learning enhanced analysis of\n EBSD data for texture representation\n\n\n\n Check for updates\n\n J. Wanni 1, C. A. Bronkhorst 1,2 & D. J. Thoma 1,2\n\n\n\n Generating reduced-order, synthetic grain structure datasets that accurately represent the measured\n grain structure of a material is important for reducing the cost and increasing the accuracy of\n computational crystal plasticity efforts. This study introduces a machine-learning-based approach,\n termed texture adaptive clustering and sampling (TACS), for generating representative Euler angle\n datasets that accurately mimic the crystallographic texture. The TACS approach employs K-means\n clustering and density-based sampling in a closed-loop iteration to create representative Euler angle1234567890():,; 1234567890():,;\n datasets. Proof-of-principle experiments were performed on rolled and recrystallized low-carbon\n steel. Validation of the TACS approach was extended to twenty-two datasets, varying lattice\n structures, and complex crystallographic textures, thereby encompassing a broad range of materials\n and crystal structures. Kolmogorov-Smirnov (K-S) test comparisons underscore the performance of\n the TACS approach over traditional electron backscatter diffraction EBSD dataset reduction\n techniques, with average K-S test scores nearing 0.9, indicating a high-\ufb01delity representation of the\n original datasets. In contrast, conventional methods display scores below 0.3, indicating less reliability\n of the structure representation. The independence of the TACS approach from material texture and its\n capability to autonomously generate datasets with predetermined data points demonstrates its\n unbiased potential in streamlining dataset preparation for crystallographic analysis.\n\n\n\n The microstructure of polycrystalline metallic materials is a very simulations are done by synthesizing a grain structure that mimics the\n important factor in determining their physical properties1\u20134. Pre- microstructural characteristics of the material. Typically, the information\n dominantly, the grain structure stands as the most observable micro- for these grain structures is initially obtained through electron backscatter\n structural feature in most polycrystalline metals, wielding a signi\ufb01cant diffraction (EBSD) datasets, which provide detailed mappings of grain\n in\ufb02uence on the structural behavior of the metal1,5,6. Speci\ufb01cally, aspects orientations. A great deal of effort is required to ensure that the character-\n such as grain shape (morphological texture) and lattice orientation istics of the synthesized grain structure accurately re\ufb02ect the actual micro-\n (crystallographic texture) can lead to mechanical property directional structure. Grain morphology is usually modeled using 3D software such as\n anisotropy5\u20138. In addition, morphological and crystallographic texture Dream3D, where the number of grains in the synthetic structure can vary\n can also in\ufb02uence the contribution of different deformation mechan- from a few tens to thousands, in\ufb02uenced by the available computational\n isms, such as dislocation slip and deformation twinning, to the overall resources and the choice of the crystal plasticity simulation approach13\u201316.\n deformation behavior of a metallic material7,8. As such, the grain However, while typical EBSD datasets for grain structures may\n structure has the potential to be engineered for enhancing speci\ufb01c encompass hundreds to thousands of grains, the grains included in a\n properties like strength and ductility7,9\u201311. Accurately de\ufb01ning the cor- representative volume element (RVE) might be reduced to manage\n relation between the grain structure of a metal and its properties is computational expenses effectively. The extent of this reduction can\n critical for developing useful computational methods for predicting and vary, depending on the speci\ufb01c requirements and objectives of the\n designing the mechanical response of materials. simulation. Such reductions necessitate meticulous attention to ensure\n Crystal plasticity simulations offer a powerful tool for studying the that the comprehensive grain information is accurately represented in\n in\ufb02uence of grain morphology and lattice orientation on the mechanical the RVE13,15\u201317,19. Given the complexities involved in capturing such\n response of crystalline materials12\u201318. Together with a theory representing detailed microstructural information within a reduced-order dataset,\n the elastic-plastic response of the single crystals, crystal mechanics maintaining the integrity of the crystallographic texture without\n\n\n\n 1Department of Materials Science and Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA. 2Department of Mechanical Engineering, Uni-\n versity of Wisconsin-Madison, Madison, WI 53706, USA. e-mail: dthoma@wisc.edu\n\n\n\n npj Computational Materials | (2024)10:133 1https://doi.org/10.1038/s41524-024-01324-4 Article\n\noverwhelming computational resources is a challenge in crystal plasti- Results and discussion\ncity simulations13,15\u201317,19. Proof-of-principle evaluation\n Generating accurate reduced-order Euler angle datasets is rele- The pole \ufb01gure (PF) maps of the EBSD dataset of a rolled and recrystallized\nvant not only for crystal plasticity simulations but also for enhancing low carbon (0.23 wt.% C) steel that was used to develop the algorithm are\nthe effectiveness of other modeling techniques such as reduced-order depicted in Fig. 1a.The rolling direction (RD) and normal direction (ND) of\nmodeling and surrogate modeling20\u201326. These methods aim to the 12.7 mm thick plate are indicated for reference. The EBSD analysis\nstreamline complex physical models, facilitating faster simulations identi\ufb01ed onlytheiron-BCCphase,resultingfromthequenchingstepinthe\nwhile maintaining high accuracy. For instance, in reduced-order manufacturing process. There is a possibility that EBSD did not detect\nmodeling, key characteristics of a material\u2019s behavior can be ef\ufb01- retained austenite or carbides in the material or mis-indexed the patterns39.\nciently encapsulated with fewer degrees of freedom, proving essential However,thisisnotrelated to thescopeof thisstudy.Astep sizeof 0.250 \u03bcm\nfor real-time simulations and iterative design processes21\u201324,26. Simi- was used to collect the data and the dataset consists of 165088 indexed\nlarly, surrogate models, which act as proxies for more intricate points. A roughly 120 \u00d7 90 \u03bcm2 region was scanned. The dataset consists of\nsimulations, can be optimized through the use of reduced-order 2068 grains, hence the dataset is large enough to represent the grain\ndatasets that capture microstructural characteristics critical to structure of the material. The pole \ufb01gure (PF) maps reveal a weak (100)\nmaterial properties20,25. As these modeling approaches are still evol- texture\u2014twice the random distribution\u2014in the rolling direction of the\nving, particularly in their capacity to predict material deformation plates. The (110) and (111) directions also show weak textures at an angle of\nbehavior, they stand to gain signi\ufb01cantly from advanced methods that 45 degrees to the RD, possibly due to the (100) texture in the RD. The weak\nef\ufb01ciently compress dataset sizes while accurately representing the texture of the EBSD dataset, characterized by more randomly oriented\nmicrostructure of the material. grains with a broad variation in orientations, poses a signi\ufb01cant challenge in\n Machine learning (ML) based approacheshave already been utilized to generating a representative reduced-order synthetic Euler angle dataset.\nsolve complex problems that are associated with EBSD datasets27\u201330. For Accurately capturing the diverse orientation characteristics of such weakly\ninstance, K. Kaufman et al. demonstrated the use of deep learning, speci- textured materials typically requires a large number of g\n\n## 2. stanek2020comparisonofsegmentation pages 2-7 (Context ID: pqac-00000007)\n\nture. by a boundary de\ufb01ned by a given threshold of misori-\nThe polycrystalline microstructures are observed by entation so that the sets are detected as two grains,\n3D EBSD and 2D EBSD. The approaches presented while in 3D case, all these points belong to one larger\nin Section 2 for simulated toy data examples are dis- set which is surrounded by a boundary. Therefore in\ncussed subsequently for both materials. The curve of the 3D case, all these points lie in the same grain,\ndependence of the mean number of grains, in a single see Fig. 1. This behaviour can be observed when the\n2D slice, on the misorientation tolerance level, is plot- grains are nonconvex, or when there is a boundary of\nted for both 2D and 3D methods. The di\ufb00erences are the given misorientation inside the grain. One could\nremarkable. conclude such a situation to be unrealistic, but the\n In Section 4, as an application to microstructure same behaviour can be observed in materials formed\nanalysis of 3D EBSD data, we present a \ufb01ve-parameter by convex grains with very low local misorientation\nanalysis of grain boundaries in Cu specimen and sug- (all voxels from a single grain have nearly the same\ngest a physical interpretation. Three levels of seg- orientation), if the chosen misorientation threshold is\nmentation are considered and discussed. The 3D mi- higher than the minimal misorientation of the neigh- J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 303\n\n\n\n\n\nFig. 2. A Laguerre tessellation simulated from the 3D\nstochastic Gibbs-Laguerre model in a cubic window. The\npattern (a) involves 1560 grains coloured at random. In\n(b), there is a 2D slice taken from the middle of the pat-\n tern (a). Fig. 3. Histograms of misorientations between neighbour-\n ing grains in the simulated data. (a) the case (A) \u2013 low\n angle grain boundaries, (b) the case (B) \u2013 high angle grain\n boundaries.\nbouring grains. In this case, the segmentation joins\nmore real grains into one larger grain. The situation\nis visualized in Fig. 1b. In both cases in the \ufb01gure, we tations of cubic crystal grid of grains were simulated,\nhave three grains which are segmented in 3D into one namely:\ngrain, because some boundaries (dashed lines) have (A) the misorientation between neighbouring\nlower misorientation than the chosen threshold, but grains is low, (see Fig. 3a),\nthe segmentation in 2D (in the cutting plane) sepa- (B) the misorientation between neighbouring\nrates these grains because the boundary in the cutting grains is high, (see Fig. 3b).\nplane (marked by the solid line) has greater misorien- These two cases represent the situations, when the\ntation than the chosen threshold. material prefers low angle grain boundaries (LAGB)\n In this section, we study the di\ufb00erence between or high angle grain boundaries (HAGB), respectively.\nsegmentation into grains in 2D and 3D for simulated For further purposes, we work with voxelized data\ndata. As a model for a polycrystalline microstructure, (100 \u00d7 100 \u00d7 100 voxels), where each voxel has its\nwe use the Laguerre tessellation [15], see Fig. 2. The orientation given by the associated grain. The data\nMarkov chain Monte Carlo method of simulation of a are processed by the software DREAM.3D. Since the\nGibbs-Laguerre tessellation [16] is extended by adding cells in the Laguerre tessellation (representing grains\nthe Euler angles to the state space. For this model of a in our case) are convex, and all voxels in each grain\ntessellation structure, two di\ufb00erent scenarios of orien- are assigned the same orientation, the simulated data304 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n\n\n\nFig. 4: The segmentation of simulated data for the case (A) in 2D with the misorientation threshold (a) 2\u25e6, (b) 2.5\u25e6, and\n(c) 3\u25e6and the same section from 3D reconstruction with the misorientation threshold (d) 2\u25e6, (e) 2.5\u25e6, and (f) 3\u25e6. The\n grains are coloured at random.\n\n\nrepresent a material without noise, i.e., all the grains into grains are considered, using the same \ufb01lters and\nare well de\ufb01ned. the same settings as in 3D. In this way, two di\ufb00erent\n To compare the results obtained in 2D and 3D, segmentations of the same slice are received, which are\nthe \ufb01lter \u2018Segment Features (Misorientation)\u2019 with a comparable. The \ufb01rst one represents the processing in\nchosen misorientation threshold for the entire 3D data 3D, while the second one represents the processing in\nset is used \ufb01rst. Then, a single slice from the processed 2D.\n3D data is cut out. This slice is treated separately, i.e., Just note that in the present study, the slice num-\nonly the voxels from the slice for the segmentation ber 50 is processed, which is shown in Fig. 2b. J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 305\n\n\n\n\n\nFig. 5. The segmentation of simulated data for the case (B) in 2D with the misorientation threshold (a) 2\u25e6, (b) 36\u25e6, and\n(c) 40\u25e6and the same section from 3D reconstruction with the misorientation threshold (d) 2\u25e6, (e) 36\u25e6, and (f) 40\u25e6. The\n grains are coloured at random.\n\n\n The di\ufb00erence between segmentation in 2D and 3D For the case (B), a similar result is shown in Fig. 5.\nin the case (A) is shown in Fig. 4. In this case, the mis- In this case, we use the misorientation thresholds equal\norientation thresholds equal to 2\u25e6, 2.5\u25e6, and 3\u25e6are used. to 2\u25e6, 35\u25e6, and 40\u25e6. Note that 0.01, 5.0, and 9.4 %\nNote that 0.15, 13.2, and 26.5 % of boundaries have of boundaries have misorientation lower than 2\u25e6, 35\u25e6,\nmisorientation lower than 2\u25e6, 2.5\u25e6, and 3\u25e6, respecti- and 40\u25e6, respectively. While we get similar segmenta-\nvely. While for the threshold of 2\u25e6, we get nearly the tions for all chosen thresholds between 2\u25e6and 35\u25e6, the\nsame segmentation in both cases, for the thresholds of di\ufb00erence between segmentations for the threshold of\n2.5\u25e6and 3\u25e6, the segmentations are entirely di\ufb00erent. 40\u25e6is not marginal.306 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n\n\n\nFig. 6. The segmentation of simulated data with the chosen misorientation threshold 2.5\u25e6for (a) 2D data, (b) 3D data \u2013\n10 slices, (c) 3D data \u2013 20 slices, (d) 3D data \u2013 30 slices, (e) 3D data \u2013 50 slices, and (f) 3D data \u2013 100 slices. The grains\n are coloured at random.\n\n\n The question is, how does the di\ufb00erence between number 50, which has already been studied. Six sce-\nsegmentation in 2D and 3D depend on the thickness of narios were compared: (a) segmentation in 2D, where\nthe specimen. Obviously, if we work with a 3D data set only the slice number 50 was used, (b) segmentation\nwhich contains only a few slices, the di\ufb00erence between in 3D with the data set of 10 slices enumerated 45\u201354,\nsegmentation in 2D and 3D should be marginal, since (c) segmentation in 3D with the data set of 20 slices\nthere is not enough space for the e\ufb00ect described in enumerated 40\u201359, (d) segmentation in 3D with the\nFig. 1. To study this dependence, the simulated data data set of 30 slices enumerated 35\u201364, (e) segmen-\n(A) were used, and the focus was given on the slice tation in 3D with the data set of 50 slices enumer- J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 307\n\n\n tion in 2D and 3D for the cases (A), (C), and (D)\n is investigated. For all three cases, the segmentation\n with three misorientation thresholds is studied. The\n \ufb01rst one is 0.1\u25e6(called 0 % misorientation segmen-\n tation in the sequel), the second one is chosen so\n that 10 % of boundaries have lower misorientation\n (called 10 % misorientation segmentation in the se-\n quel). The last one corresponds to 15 % of boundaries\n with misorientation lo\n\n## 3. stanek2020comparisonofsegmentation pages 1-2 (Context ID: pqac-00000008)\n\n Kovove Mater. 58 2020 301\u2013319 301\n DOI: 10.4149/km 2020 5 301\n\n\n\n Comparison of segmentation of 2D and 3D EBSD measurements\n in polycrystalline materials\n\n\n J. Stan\u02c7ek1*, J. Kope\u010dek2, P. Kr\u00e1l3, I. Kara\ufb01\u00e1tov\u00e14, F. Seitl4, V. Bene\u01614\n\n\n\n 1Department of Mathematics Education, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n 2Institute of Physics of the CAS, Department of Functional Materials, Na Slovance 2, 182 00 Prague, Czech Republic\n 3Institute of Physics of Materials, ASCR, \u02c7Zi\u017ekova 22, 616 62 Brno, Czech Republic\n 4Department of Probability and Mathematical Statistics, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n\n\n Received 30 December 2019, received in revised form 26 April 2020, accepted 17 June 2020\n\n\n Abstract\n\n A comparison study of grain segmentation in 2D and 3D electron backscatter di\ufb00rac-\n tion (EBSD) measurements of polycrystalline materials, based on a misorientation threshold,\n is presented. Using Laguerre tessellations with orientation marks, a simulation study that\n demonstrates a di\ufb00erent sensitivity of segmentation in the case of low angle and high angle\n grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investi-\n gated concerning segmentation, and there are given some recommendations for the misorien-\n tation thresholds. Computational tools such as DREAM 3D software enable signi\ufb01cant data\n evaluation that is completed by a \ufb01ve-parameter analysis of grain boundaries.\n\n\n K e y w o r d s: electron backscatter di\ufb00raction (EBSD), grain boundaries, misorientation,\n tessellation\n\n\n\n 1. Introduction observed the boundaries \u2013 the separators, whereas\n grains or rather their interiors were homogeneous ar-\n The electron backscattered di\ufb00raction method eas inside with de\ufb01ned or expected value of properties.\n(EBSD) has been a powerful tool in material science Of course, it is known that a grain structure with the\nsince the nineties, when a certain level of computeri- tessellation formula can be created [7]; nevertheless,\nzation and application of Hough transform took place, EBSD is the \ufb01nal step reversing axiom to construct\nsee [1, 2]. Nowadays, we can perform not only 2D the grains [2]. We have a set of measured points in the\nEBSD but exploiting focused ion beams (FIB), even EBSD method. We have positions of such points and\n3D EBSD, by materials\u2019 slicing tomography [3]. The their marks (e.g., Euler angles). Thus, here the inte-\n3D EBSD data evaluation is a complicated process [4]. rior is a quality, which exists, whereas the boundary\nStill, we expect to be able to construct grains [5] and has to be constructed. The easiest way how to create\nmicrostructure parameters by applying the same or grains from the set of measured points is the following\nsimilar procedures as in 2D EBSD data evaluation [6]. de\ufb01nition: the grain interior shall have the same or\nHowever, a discrepancy in the segmentation of data very similar properties. Thus, we construct grains by\npoints into the grains in the 2D and 3D cases is ap- collecting the measured points with the de\ufb01ned prop-\nparent, since a nonconvex grain in 3D can be observed erty. The metallographic and EBSD grains must be\nas more than one grain in a 2D slice. then the same in the optimal case. However, the re-\n The segmentation problem is deeply connected to sult of an experiment is not optimal. It contains some\nthe basic understanding of the polycrystals. Gener- noise or even a kind of systematic aberrations, and the\nally, the de\ufb01nition of grains as they were observed by reconstruction of grain structure can be done only up\nfounders of metallography was purely natural: They to some level of accuracy.\n\n*Corresponding author: tel.: +420 951553215; e-mail address: stanekj@karlin.m.cuni.cz302 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n The problem described above was omitted in com-\nputational materials science since the 3D data is fre-\nquently used in an elementary manner. Despite the\nexplosion of three-dimensional methods in the last\ndecades, the majority of 3D models use simple math-\nematical schemes to build the structure by a general\nprocedure approaching the reality by just a few pa-\nrameters, measured from 2D cuts of material [8].\n Data obtained by 3D EBSD or 2D EBSD form a\nset of orientations of measured points in a 3D ma-\ntrix (the points are called voxels in the sequel) or\n2D matrix (pixels in the sequel), respectively. The\nEBSD method considers a grain to be the set of all\npoints (voxels or pixels, respectively) with a similar\norientation, which are surrounded by a boundary of\na given misorientation threshold. The typical thresh-\nold in the EDAX OIM software is 5\u25e6[9], but other\nchoices are possible as well [10, 11]. The obtained seg-\nmentation into the grains depends on the chosen mis- Fig. 1. Di\ufb00erences between slices processed from 2D EBSD\n and the same slice processed from 3D EBSD. The greyorientation threshold, of course, but even if we use a\n colour displays the real grains (in the plane, which is or-\n\ufb01xed misorientation threshold, we may get di\ufb00erent\n thogonal to the cutting plane). The blue colour displays\nresults of the segmentation in 2D and 3D, as shown their 3D segmentation in the cutting plane, i.e., the parts\nin [12]. Therefore, the results obtained in 2D EBSD are considered to belong to the same grain. Finally, the\nmay not be easily compared with the results from 3D red and the orange colours show their 2D segmentation\nEBSD. in which they are considered to be di\ufb00erent grains. (a)\n In the paper, we present a comparison study of The situation with a nonconvex grain or a grain with a\nsegmentation of 2D and 3D EBSD measurements in boundary inside. (b) The situation with convex grains,\n boundaries with the misorientation smaller than the cho-polycrystalline materials, based on a misorientation\n sen threshold are denoted by the dashed line.\nthreshold. The main results of this topic are presented\nin Sections 2 and 3. The aim is to draw the reader\u2019s\nattention to the danger of obtaining misleading quan-\ntitative results about the geometry of microstructure crostructure analyses make use of DREAM 3D soft-\nwhen paying little attention to a proper segmentation. ware [14].\nFirst, in Section 2, the problem is demonstrated in a Some conclusions from Sections 2 and 3 are given\nsimulation study. A generalized Laguerre tessellation in more detail in the \ufb01nal part of the paper.\nmodeling of polycrystalline microstructure measured\nby 3D EBSD was used in [13]. We can simulate this\nmodel with a priori given distribution of misorienta- 2. Di\ufb00erences between 2D and 3D\ntions. The in\ufb02uence of this distribution on the seg- segmentation\nmentation is shown.\n In Section 3, we discuss the segmentation prob- The di\ufb00erences between segmentation into grains\nlem with two materials, namely the AlMgSc alloy af- in 2D and 3D are caused by the fact that in a 2D\nter ECAP and annealing, and pure copper after high- cutting plane, two sets of voxels may be surrounded\npressure torsion and annealing at room temperature. by a boundary de\ufb01ned by a given threshold of misori-\nThe polycrystalline microstructures are observed by entation so that the sets are detected as two grains,\n3D EBSD and 2D EBSD. The approaches presented while in 3D case, all these points belong to one larger\nin Section 2 for simulated toy data examples are dis- set which is surrounded by a boundary. Therefore in\ncussed subsequently for both materials. The curve of the 3D case, all these points lie in the same grain,\ndependence of the mean number of grains, in a single see Fig. 1. This behaviour can be observed when the\n2D slice, on the misorientation tolerance level, is plot- grains are nonconvex, or when there is a boundary of\nted for both 2D and 3D meth\n\n## 4. savage2021anautomatedprocedure pages 1-2 (Context ID: pqac-00000009)\n\n Materials Characterization 171 (2021) 110808\n\n Contents lists available at ScienceDirect\n Materials Characterization\n\n\n journal homepage: www.elsevier.com/locate/matchar\n\n\nAn automated procedure built on MTEX for reconstructing deformation\ntwin hierarchies from electron backscattered diffraction datasets of heavily\ntwinned microstructures\nDaniel J. Savage a,b, Rodney J. McCabe b, Marko Knezevic a,*\na Department of Mechanical Engineering, University of New Hampshire, Durham, NH 03824, USA\nb Materials Science and Technology Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA\n\n\nA R T I C L E I N F O A B S T R A C T\n\nKeywords: This paper presents a set of algorithms built on the MTEX and MATLAB graph toolboxes for automatic recon\u00ad\nMicrostructures struction of deformation twin hierarchies from Electron Backscatter Diffraction (EBSD) datasets with a focus on\nTwinning developing methods for heavily twinned microstructures (twin fractions >0.5). The algorithms address key issues\nTitanium arising at large strains, mainly: missing twin relationships, grouping of heavily deformed grain fragments intoEBSD families of similar orientation originating from a single initial grain, identification of parent fragments for largeMTEX\n twin volume fractions, and classification of families having twin relationships with multiple families. To facilitate\n the development of these algorithms, large-grained ultra-high purity \u03b1-Ti deformed in compression along two\n directions is investigated. Graphs are utilized to handle non-local geometric merging and to represent re\u00ad\n lationships throughout the reconstruction process. When determining if a grain fragment is from the undeformed\n microstructure, the combined metrics of the fragment\u2019s orientation volume fraction in the initial texture and the\n directed graph centrality measure of out-closeness (the number of nodes reached in a graph from a given node)\n are essential. To address automation in reconstructing the sequence of twinning and relating fragments origi\u00ad\n nating from a single grain in the initial microstructure, the twin family tree is formulated as a minimum spanning\n tree emanating from the initial grain family. A scheme constructing the distances associated with twin rela\u00ad\n tionship comprising the spanning tree is developed, and a novel quasi-directional Prim spanning tree algorithm is\n used to determine the twin family tree. The procedure is demonstrated to significantly improve the level of\n automation in reconstructing twin hierarchies in heavily twinned microstructure compared to other methodol\u00ad\n ogies in literature. The procedure can readily be applied to analyses of twinning in metals, as well as provide an\n approach for routinely extracting twin statistics at larger deformation levels than previously possible. Signifi\u00ad\n cantly, the procedure is demonstrated to be capable of identifying third generation twinning in \u03b1-Ti\n microstructures.\n\n\n1. Introduction formation in the vicinity of their boundaries [9\u201315]. This correlation has\n been rationalized to arise due to substantial crystal reorientations and\n Deformation twinning is a shear-based transformation that results in underlying shifts in active crystallographic slip modes within the twin\u00ad\ncrystal reorientation and significantly influences the evolution of local ned domain. Furthermore, collection of nucleation statistics such as in\nand overall material properties of many important metals undergoing fatigue loading, is vital to building the understanding of fracture in\ntwinning [1\u20133]. One well-known twin phenomena is the orientation hexagonal close-packed (HCP) materials [16]. Thus, measuring and\ndependent anisotropic hardening introduced by twin lamella frag\u00ad classifying twinning is relevant for developing a working knowledge of\nmenting slip domains and reorienting towards hard crystal orientations fracture and the anisotropic response of materials.\n[4\u20138]. In addition to hardening, many studies show the formation of Twinning plays an important role in developing Integrated Compu\u00ad\ntwins are correlated to flow localization and ultimately void/crack tational Materials Engineering (ICME) toolsets. Crystal plasticity\n\n * Corresponding author at: University of New Hampshire, Department of Mechanical Engineering, 33 Academic Way, Kingsbury Hall, W119, Durham, NH 03824,\nUSA.\n E-mail address: marko.knezevic@unh.edu (M. Knezevic).\nhttps://doi.org/10.1016/j.matchar.2020.110808\nReceived 3 September 2020; Received in revised form 17 November 2020; Accepted 24 November 2020\nAvailable online 4 December 2020\n1044-5803/\u00a9 2020 Elsevier Inc. All rights reserved.D.J. Savage et al. Materials Characterization 171 (2021) 110808\n\n\n\n\n\nFig. 1. Schematic of the tasks and challenges performed in twin codes (from left to right): identification of twin relationships and grouping of grain fragments\n(clusters), identification of fragments of the same orientation (families, F), and the determination of the representative family tree. Note that a cluster is the group of\nfamilies comprising a single grain in the initial microstructure. Shown is an example of the family tree where the F1 family is the parent of families F2 and F4, and F3\nis a second generation twin whose parent is F2. A circular twin relationship (e.g. F1, F2, F3, and F1) arises from the combination of twin crystallographic reor\u00ad\nientations and a twin code must identify the correct twin hierarchy. The circular relationship causes no root (the family that is not a twin) to be evident based on the\ndirections of twin relationships, and another criterion (presented in this work) is needed to determine the root automatically.\n\nconstitutive laws, for instance, enable ICME [17\u201325] by modeling strain root by area majority rule, and relying heavily on fragment geometry to\npath dependent deformation behavior of highly anisotropic metals determine the twin. To address fundamental issues regarding post-\nthrough the evolution of slip, twin, and phase-induced texture evolution mortem analysis and slip, inverse modeling of the twinning process (e.\n[26,27]. In HCP and lower symmetry metals such as orthorhombic g. through physically based crystal plasticity modeling) in addition to\nmetals [28\u201333], it is usually necessary to emulate the twinning process, traditional twin identification approaches are needed. Key to this com\u00ad\nreproducing the statistics of experimental data to capture the mechan\u00ad bined approach is a highly flexible framework that has the capability to\nical response during complex loading [27,34,35]. The emulation of the assimilate all levels of microstructure and modeling information into a\ntwinning process requires detailed knowledge of the twin types, twin coherent interpretation of the microstructure evolution.\nfractions, twin nucleation criterion, and their role in hardening the Volume fractions of twins in highly deformed microstructures have\nmaterial; and ICME relies on the ability to extract twin-based been determined using boundary length fractions and texture based\ninformation. methods. The boundary length fraction method is based on measuring\n Considering the importance of twinning in fracture and under\u00ad the fraction of high-angle boundaries in the microstructure consistent\nstanding the anisotropic response of materials, several automatic twin with the twin misorientation relationship. The boundary fraction\nidentification\n\n## 5. polonsky2019threedimensionalanalysisand pages 2-5 (Context ID: pqac-00000010)\n\nAM.3D [8] plugin was\ntity, and resolution sufficient to make such measurements developed for BisQue in order to incorporate the materials\n[11]. Even in 2D EBSD analysis, such standards are rare science based 3D dataset analysis tools. The database\nand do not necessarily address additional subtleties arising architecture of BisQue is such that every processing\nfrom more complex, albeit commonly studied, microstruc- execution, dataset (input and output), and workflow\ntures, including twinned materials, nanocrystalline mate- processing pipeline receives a unique web address URL.\nrials, and deformed materials [12\u201317]. Standardization of These URLs are useful for at least two reasons: tracking\ncharacterization methods becomes critical as the complex- data provenance and data sharing. Data provenance with\nity of the analyses increases. For example, high-resolution BisQue is automatically captured and can be easily accessed\nEBSD (HR-EBSD) and Heaviside Digital Image Corre- through visualized workflows, as shown in Fig. 1. The\nlation (HDIC) [18] are extensions of conventional EBSD workflow shows where input data is located, the parameters\nand digital image correlation (DIC) that require specialized of a module execution are captured in the metadata, and\npost-processing to measure elastic and plastic strain at a the output data location is tracked. Therefore, passing\nresolution of 10-4 or smaller [19, 20] and 1.5\u00d710\u22123 [21], data between research colleagues becomes straightforward\nrespectively. The variety of analytic approaches available to due to the data operation transparency. Furthermore, data\nmake accurate strain measurements, as well as the sensi- sharing is facilitated due to the data all being stored\ntivity of these measurements to instrumentation and sample and accessed remotely, mitigating versioning problems\npreparation, motivates more rigorous standardization of the common to large data operations. For example, two\ntechnique via sample reference standards [22, 23]. groups collaborating between the US and Europe may findIntegr Mater Manuf Innov (2019) 8:37\u201351 39\n\n\n\n\n\nFig. 1 BisQue data provenance tree from an additively manufactured 304L Stainless Steel 3D dataset. Arrows show workflow between input data,\nDREAM.3D pipelines, module executions (mex), and associated datasets\n\n\n\ndifficulty operating on the same dataset version, especially In addition, a volume viewer enables web-based visual-\nwhen business hours do not overlap, without extensive ization of cell-level data, removing the need for programs\ncommunication. like HDFView or ParaView and their associated hardware\n requirements [31, 32]. 3D volumes collected via TriBeam\nBisQue DREAM.3D Integration tomography and reconstructed in DREAM.3D generally\n require many tens of gigabytes of RAM for visualiza-\nA DREAM.3D module was incorporated into BisQue to tion in ParaView, limiting analysis to powerful workstation\nallow for the processing and treatment of 3D datasets computers.\nwith the advantages of cloud-based infrastructure described\nin the previous section (\u201cBisQue for Cloud Based Data Materials and Data Collection\nStorage and Analysis\u201d). The DREAM.3D processing capa-\nbilities were also extended to allow for parallelized pro- The 3D datasets were collected via TriBeam tomography\ncessing across wide parameter spaces. Anytime a process- from two additively manufactured alloys in their as-\ning pipeline is generated in DREAM.3D through BisQue, built state, meaning no additional heat treatments were\nparameter input variables can be represented as a range performed on the samples after additive processing.\nof values (with sub-increments), rather than as a distinct TriBeam tomography [33] utilizes a femtosecond laser\nvalue. An example of DREAM.3D pipeline execution is for micromachining within a dual-beam scanning electron\nshown in Fig. 2 alongside the analogous representation microscope (SEM) equipped with a focused ion beam\nin DREAM.3D. The user selects an input source and (FIB). The ultrashort pulse length minimizes damage to\npipeline source, parameterizable values are then automati- the machined surface while enabling high material removal\ncally displayed in the parameter section. Using straightfor- rates, roughly four to five orders of magnitude faster than\nward list comprehension, many independent instantiations FIB milling alone, resulting in an accessible volume on\nof DREAM.3D can be run at once across multiple nodes on the order of a cubic millimeter [30, 34]. The FIB can\na compute cluster, dramatically decreasing setup and analy- also be used at a glancing angle to minimize surface\nsis time for detailed parameter studies. The resulting output roughness due to laser-induced periodic surface structures\nfiles can then be examined as individual pipelines as BisQue (LIPSS); however, this cleaning step is material dependent\nhas a fully integrated table viewer that allows for both table and is not always required to obtain high-quality 3D EBSD\nviewing and inspection of the tree structure of HDF files. data.40 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\n\n\n\nFig. 2 DREAM.3D pipeline for reconstruction of an additively manu- parameterization for variable inputs in a DREAM.3D pipeline, which\nfactured 304L Stainless Steel 3D dataset as viewed in (a) DREAM.3D can be executed in parallel on high-performance compute clusters and\non a workstation computer and (b) the DREAM.3D BisQue mod- then analyzed to determine the ideal reconstruction parameters\nule in a web browser. The DREAM.3D BisQue module allows for\n\n\n\nLENS-Processed 304L Stainless Steel EBSD scans were collected on each slice with an in-plane\n resolution of 3 \u03bcm, yielding square voxels. Roughly 200 GB\nLaser engineered net shaping (LENS) is a directed energy of data was collected from a total of 80 slices over the course\ndeposition (DED) additive manufacturing technique derived of two days, resulting in a volume of 816 \u00d7 834 \u00d7 240 \u03bcm.\nfrom laser cladding processes, in which metal powder Raw EBSD patterns, comprising nearly 90% of the collected\nis blown into a molten pool created using a laser. The data, were saved for post-processing. Individual slices were\nlow dilution area and high deposition rates achievable in reconstructed into a coherent volume using DREAM.3D\nLENS, up to an order of magnitude higher than powder- in BisQue. A minimum confidence index (CI) of 0.1 was\nbed additive processes, allow for the manufacture of large chosen as the threshold for data quality, as the indexing\nscale parts [35]. A sample of 304L stainless steel fabricated of EBSD patterns has been shown to be 95% accurate at\nby the LENS process was characterized in the TriBeam a CI of 0.1 [36]. This threshold is then used to create a\nsystem using a slice thickness of 3 \u03bcm. The femtosecond \u201cmask\u201d that flags individual voxels as containing good or\nlaser ablated surface was cleaned using glancing-angle bad (noisy) data, so the mask represents a subset of the\nFIB milling to increase the quality of EBSD patterns. total volume characterized during data collection. The maskIntegr Mater Manuf Innov (2019) 8:37\u201351 41\n\n\nfor the sample of 304L can be seen in Fig. 3. Grains were were saved for post-processing. Glancing-angle FIB was not\nsegmented using a misorientation tolerance as described required to resolve EBSD patterns, which were collected\nin \u201cMisorientation Tolerance Parameterization for Feature with an in-plane resolution of 1.5 \u03bcm to create square\nIdentification,\u201d and grains containing less than 64 voxels voxels. As with the 304L sample, a minimum CI of 0.1\nwere considered too small to be resolved and were removed. was chosen for the initial mask. Grains were segmented\nThis corresponds to 4 \u00d7 4 \u00d7 4 points across a feature and an using a misorientation tolerance of 2\u25e6using the same\nequivalent diameter of 14.9 \u03bcm, roughly five times smaller segmentation algorithm as explained in \u201cMisorientation\nthan the average grain size determined from 2D analysis Tolerance Parameterization for Feature Identification.\u201d A\ndone prior to 3D characterization. Gaps remaining from minimum size requirement of 125 voxels was applied to\nthese removed features were filled via iterative dilation of delete grai\n\n## 6. polonsky2019threedimensionalanalysisand pages 1-2 (Context ID: pqac-00000011)\n\nIntegrating Materials and Manufacturing Innovation (2019) 8:37\u201351\nhttps://doi.org/10.1007/s40192-019-00126-7\n\n TECHNICAL ARTICLE\n\n\nThree-dimensional Analysis and Reconstruction of Additively\nManufactured Materials in the Cloud-Based BisQue Infrastructure\n\nAndrew T. Polonsky1 \u00b7 Christian A. Lang2 \u00b7 Kristian G. Kvilekval2 \u00b7 Marat I. Latypov1,3 \u00b7 McLean P. Echlin1 \u00b7\nB. S. Manjunath2 \u00b7 Tresa M. Pollock1\n\n\nReceived: 30 December 2018 / Accepted: 18 February 2019 / Published online: 22 March 2019\n\u00a9 The Minerals, Metals & Materials Society 2019\n\nAbstract\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a\ncloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline\nvariables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-\ncollected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore,\na comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using\nBisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects\nfor impact in material science are also discussed.\n\nKeywords Tomography \u00b7 3D materials science \u00b7 Microstructure analytics \u00b7 Cloud infrastructure \u00b7 Additive manufacturing \u00b7\n3D metal printing\n\n\n Introduction\n Andrew T. Polonsky\n polonsky@ucsb.edu\n Recent external pressures mandated by government grants\n Christian A. Lang requiring data be accessible to the public, and internal\n clang@ucsb.edu pressures within academic communities for data sharing\n and reproducibility have led to the emergence of data-\n Kristian G. Kvilekval\n driven platforms within the scientific community [1\u20134].\n kgk@ucsb.edu\n Requirements for these platforms include providing data\n Marat I. Latypov provenance, data sharing, and hosting services for data that\n latmarat@ucsb.edu may otherwise disappear when a lab closes or hardware\n fails [5]. Recently, data repositories have shifted from\n McLean P. Echlin\n mechlin@ucsb.edu simply siloing data to include data analytics and predictive\n capabilities. Furthermore, some of these repositories are\n B. S. Manjunath directly linked with high-performance computing in order\n manj@ucsb.edu\n to populate an ever growing database, such as with\n Tresa M. Pollock thermodynamic databases such as the Materials Project [6]\n tresap@ucsb.edu and the Open Quantum Materials Database (OQMD) [7].\n More sophisticated predictive/interactive databases, such as\n1 the aforementioned thermodynamic ones, have been slower Materials Department, University of California Santa Barbara,\n Santa Barbara, CA 93106, USA to develop with the 3D Materials Science community. This\n2 is likely due to the less standardized and more diverse data Electrical and Computer Engineering Department,\n University of California Santa Barbara, Santa Barbara, modalities, the complexity and diversity of microstructures,\n CA 93106, USA and the relative infancy of the community itself.\n3 Mechanical Engineering Department, University of California In almost all regards, using 3D data to answer scientific\n Santa Barbara, Santa Barbara, CA 93106, USA questions in Materials Science is challenging due to the38 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\nneed for dataset acquisition, reconstruction, and analysis. In 3D characterization techniques, such as X-ray-based\nReconstruction of 3D datasets is a complex process with diffraction contrast tomography (DCT), the problems of\nmany degrees of freedom. The algorithms employed, and data reconstruction and analysis are also convoluted. Coher-\ntolerances selected by the user to align slices, segment ent volumes must be reconstructed from individual 2D\nfeatures, and remove data artifacts (i.e., remove noise), images via back projection techniques [24]. The diffraction\nultimately determine the quantitative data extracted from spots from the 2D projections are generally assumed to orig-\nthe volume. Significant progress on the reconstruction and inate from a single grain, limiting the characterization of\nanalysis of electron backscatter diffraction (EBSD) data has materials with large internal strain gradients [25]. Moreover,\nbeen made by a software package DREAM.3D [8], which stereological conventions established in 2D do not neces-\ncreates a unified data structure for 3D data treatment and sarily capture the real 3D microstructures, particularly at\naggregates tools and algorithms collectively developed by extrema of property distributions [26, 27]. Given the wealth\nthe community. In DREAM.3D, this process takes the form of additional data provided by 3D characterization, new\nof a pipeline, which typically consists of well over a dozen tools aimed at exploring the parameter space offered to 3D\nsteps, and in which each step has its own set of input reconstruction are required. In the following manuscript, we\nvariables and outputs. DREAM.3D typically runs in a serial describe the use of a cloud-based platform, BisQue [28, 29],\nprocess on a workstation, with predefined input parameters. enhanced with the DREAM.3D microstructural software\nDue to the long computation times required to process analysis package for the 3D reconstruction and analysis\n3D data, which often scales proportionately as dataset size of additively manufactured microstructures characterized\nand complexity increases, it is impractical to explore the using TriBeam tomography [30].\nfull parameter space by manually running reconstructions.\nTypically, only several variables are adjusted by a user until\na qualitatively acceptable reconstruction is attained. This Methods\nprocess inherently imparts bias and typically provides little\nopportunity to justify specific reconstruction variables. The following sections describe the software infrastructure\n The bias generated during reconstruction of 3D datasets and materials investigated.\nis subsequently carried through to data analysis, which,\nfor microstructural data, is still an ill-defined process. BisQue for Cloud-Based Data Storage and Analysis\nStandardization of analysis involving new characterization\ntechniques is generally a slow process. Automated orien- All data processing was performed in the cloud with BisQue\ntation imaging analysis via EBSD first emerged in the [28, 29] using a web browser interface to interact with the\nearly 90s, but standards did not emerge for nearly two 3D datasets. BisQue has backend infrastructure support for\ndecades [9, 10]. These new standards covered basic quan- web services such as Amazon S3, IRODS, and federated\ntitative analysis, including methods for grain size measure- local file systems as well as client facing tools such as\nment, as well as guidelines governing data quality, quan- Java, Python, and Matlab. A DREAM.3D [8] plugin was\ntity, and resolution sufficient to make such measurements developed for BisQue in order to incorporate the materials\n[11]. Even in 2D EBSD analysis, such standards are rare science based 3D dataset analysis tools. The database\nand do not necessarily address additional subtleties arising architecture of BisQue is such that every processing\nfrom more complex, albeit commonly studied, microstruc- execution, dataset (input and output), and workflow\ntures, including twinned materials, nanocrystalline mate- processing pipeline receives a unique web address URL.\nrials, and deformed materials [12\u201317]. Standardization of These URLs are useful for at least two reasons: tracking\ncharacterization meth\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011]\n\nStatus: Paper Count=22 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=12 | Current Cost=$0.1324 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_1oiOo0bOuigk4ZUPXbdj7xZ2", + "info": { + "start_ts": 805.142754992, + "end_ts": 819.6230742 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_eVCSSI1d7fc6pWTEoIMUSTF0", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 1, \"updated_plan_objective_json\": \"{\\\"id\\\":1,\\\"objective\\\":\\\"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\\\",\\\"rationale\\\":\\\"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\\\",\\\"evaluation\\\":\\\"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_eVCSSI1d7fc6pWTEoIMUSTF0", + "info": { + "start_ts": 845.159035053, + "end_ts": 845.160391077 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_UpEp4rBNUwkVd82OjFFGqJTn", + "type": "function", + "function": { + "arguments": "{\"question\": \"What peer-reviewed evidence establishes MTEX as a benchmark for EBSD grain detection/reconstruction and crystallographic texture analysis, including ODF, pole figures, IPF, and advanced twin reconstruction? What cautions apply to misorientation-threshold grain segmentation?\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Most relevant evidence (with ids):\n\n## pqac-00000012\n\nFrom text 'savage2021anautomatedprocedure pages 1-2'.\n\nDaniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\nThis peer-reviewed paper (Materials Characterization, 2021) demonstrates MTEX's capability for advanced EBSD analysis, specifically automated reconstruction of deformation twin hierarchies in heavily twinned microstructures. The work addresses critical challenges in twin identification including missing twin relationships, grouping of deformed grain fragments, and classification of multi-family twin relationships. The authors developed algorithms combining MTEX with MATLAB graph toolboxes to significantly improve automation compared to existing methodologies. The procedure successfully identifies third-generation twinning in \u03b1-titanium and can extract twin statistics at larger deformation levels than previously possible. However, the excerpt does not provide information about MTEX's performance regarding ODF, pole figures, or IPF analysis, nor does it discuss cautions regarding misorientation-threshold grain segmentation methods.\n\n\n---\n\n## pqac-00000013\n\nFrom text 'savage2021anautomatedprocedure pages 3-4'.\n\nDaniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\nThis peer-reviewed article from Materials Characterization demonstrates MTEX's application for grain reconstruction and texture analysis in heavily twinned \u03b1-Ti microstructures. The authors use MTEX's texture analysis capabilities to handle complex microstructures with large twin volume fractions, multiple twin variants, and higher-order twinning. The paper presents inverse pole figure (IPF) EBSD maps and discusses texture analysis with strong c-axis texture observations. Regarding misorientation-threshold grain segmentation cautions, the authors note that due to strong texture, neighboring grains often have low-angle boundaries, and twins commonly chain across grain boundaries with only small misorientation differences. They address this by biasing reconstruction towards over-segmentation rather than under-segmentation and using different misorientation tolerances (1-2\u00b0) depending on loading direction. The authors acknowledge that under-segmentation and low-angle boundary treatment presents challenges in grain reconstruction.\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013]\n\nStatus: Paper Count=22 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=14 | Current Cost=$0.1716 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_UpEp4rBNUwkVd82OjFFGqJTn", + "info": { + "start_ts": 868.044797633, + "end_ts": 874.62140145 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_P8jQvXEYp1OGYeb95uE317Pf", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 2, \"updated_plan_objective_json\": \"{\\\"id\\\":2,\\\"objective\\\":\\\"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\\\",\\\"rationale\\\":\\\"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\\\",\\\"evaluation\\\":\\\"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_P8jQvXEYp1OGYeb95uE317Pf", + "info": { + "start_ts": 898.53099696, + "end_ts": 898.532407154 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_QJeM4kTCkLbZRr827PUgMWr1", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 3, \"updated_plan_objective_json\": \"{\\\"id\\\":3,\\\"objective\\\":\\\"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\\\",\\\"rationale\\\":\\\"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\\\",\\\"evaluation\\\":\\\"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_QJeM4kTCkLbZRr827PUgMWr1", + "info": { + "start_ts": 923.49329889, + "end_ts": 923.494253073 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_ibuikCX928XmGVA9Di78lliY", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 4, \"updated_plan_objective_json\": \"{\\\"id\\\":4,\\\"objective\\\":\\\"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\\\",\\\"rationale\\\":\\\"The answer should support a concrete implementation decision, not merely list software.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\\\",\\\"evaluation\\\":\\\"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"completed\",\n \"result\": \"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\",\n \"evaluation\": \"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_ibuikCX928XmGVA9Di78lliY", + "info": { + "start_ts": 957.818000802, + "end_ts": 957.820227968 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_FRljLsn8HFS1KSTBdRAWzMFk", + "type": "function", + "function": { + "arguments": "{\"artifact_type\": \"table\", \"description\": \"Create a comparison matrix for MTEX, kikuchipy, orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, and OpenXY/research scripts. Columns: package/ecosystem; Python-native status; grain reconstruction/segmentation; texture ODF/PF/IPF; KAM/GND; maturity/reference status; recommended role. Use concise capability ratings such as strong, partial/custom, absent/not core, and clearly avoid claiming kikuchipy or pyxem are complete MTEX replacements.\"}", + "name": "create_artifact" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "### Artifact (ID artifact-00)\n\n| Package / ecosystem | Python-native status | Grain reconstruction / segmentation | Texture ODF / PF / IPF | KAM / GND | Maturity / reference status | Recommended role |\n|---|---|---|---|---|---|---|\n| MTEX | No; MATLAB-based | **Strong** for symmetry-aware 2D grain detection/reconstruction; advanced specialized workflows documented | **Strong**; benchmark for ODF, pole figures, IPF, texture components (pqac-00000000, pqac-00000012, pqac-00000013) | **Strong** in the MTEX ecosystem; widely treated as benchmark | Very mature, heavily cited benchmark; peer-reviewed advanced EBSD workflows documented (pqac-00000012, pqac-00000013) | Functional benchmark and validation target, but not MATLAB-free |\n| kikuchipy | Yes | **Absent / not core** for downstream grain reconstruction; primarily EBSD pattern processing/indexing support | **Partial / custom** via interoperability, not a full texture-analysis suite | **Absent / not core** | Open-source Python EBSD toolbox with literature record, but not a complete post-processing replacement in the retrieved evidence | Upstream pattern handling, preprocessing, indexing, data I/O; pair with orix and/or DREAM3D-NX |\n| orix | Yes | **Partial / custom**; orientation objects and misorientation logic support custom grain segmentation, but not MTEX-level turnkey reconstruction | **Strong / near-benchmark for orientation visualization**; IPF/PF native, ODF possible but typically less turnkey than MTEX | **Partial / custom**; KAM-like metrics and curvature workflows need user implementation/validation | Peer-reviewed library record available; positioned as crystallographic orientation library rather than full EBSD suite | Core Python orientation engine for custom EBSD post-processing and visualization |\n| DefDAP | Yes | **Partial**; useful 2D deformation/EBSD map grain analysis, but narrower than MTEX | **Partial**; some crystallographic plotting/analysis, not a full MTEX texture environment | **Partial**; deformation-oriented metrics are a strength, GND generally custom/limited | Open-source Python project with software-release citation; maturity moderate, community smaller than MTEX/DREAM3D | Good lightweight Python choice for 2D grain/deformation analysis after indexing |\n| DREAM3D-NX / simplnx | Yes (Python API around DREAM3D-NX) | **Strong**; mature pipeline-based segmentation/reconstruction, especially for 2D/3D EBSD and batch automation (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000011) | **Partial**; some crystallographic statistics, but not as texture-centric or interactive as MTEX (pqac-00000004) | **Partial / custom**; not typically the first choice for KAM/GND-centric analysis | Mature EBSD/microstructure pipeline ecosystem with peer-reviewed reconstruction studies (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Best Python-accessible option for robust automated grain segmentation/statistics in a MATLAB-free pipeline |\n| pyxem | Yes | **Absent / not core for SEM-EBSD**; focused on diffraction microscopy in TEM/4D-STEM, not a direct MTEX replacement | **Partial** for orientation mapping/visualization in diffraction-microscopy contexts | **Absent / not core** | Mature open-source microscopy library, but its center of gravity is 4D-STEM/SPED rather than conventional SEM-EBSD post-processing | Use only if your workflow spans diffraction microscopy; not recommended as primary EBSD grain-texture backend |\n| PyMicro | Yes | **Partial / custom**; useful microstructure data structures and crystallography, but less turnkey for EBSD grain reconstruction | **Partial / custom** | **Partial / custom** | Useful research software ecosystem, but less established as a mainstream EBSD post-processing stack in the retrieved evidence | Supporting library for custom workflows, data models, simulation coupling |\n| OpenXY / research scripts / pyEBSD-style code | Yes, but heterogeneous | **Partial / custom** to **strong in isolated tasks**; highly variable and often method-specific | **Partial / custom** | **Partial / custom**, including possible GND research implementations | Lowest standardization; often lacks broad peer-reviewed software validation or stable APIs | Fill specific gaps such as KAM/GND or custom segmentation, but validate carefully against MTEX/DREAM3D outputs |\n| Practical no-MATLAB stack for AM AlSi10Mg | Yes | **Strong overall** if composed as: kikuchipy/PyEBSDIndex + DREAM3D-NX for segmentation + orix/DefDAP for orientation/statistics | **Partial-to-strong overall**; strongest with orix plus custom ODF tooling | **Partial / custom overall**; KAM feasible, GND should be a separately validated module | Most realistic MATLAB-free path based on current open-source ecosystem and evidence balance (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Recommended integrated pipeline rather than searching for a single Python package that fully replaces MTEX |\n\n\n*Table: This table compares major open-source Python-native or Python-accessible EBSD post-processing options against MTEX for grain reconstruction, texture analysis, and misorientation-derived metrics. It is designed to help choose a MATLAB-free downstream pipeline for automated EBSD workflows.*\n\nArtifact IDs that may be injected into the answer: [artifact-00]\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013]\n\nStatus: Paper Count=22 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=14 | Current Cost=$0.1716 | Disease-Target Evidence=0", + "name": "create_artifact", + "tool_call_id": "call_FRljLsn8HFS1KSTBdRAWzMFk", + "info": { + "start_ts": 980.285662375, + "end_ts": 993.711719606 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_7bcoFGw2YuGzai5ip8f8B7P4", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 5, \"updated_plan_objective_json\": \"{\\\"id\\\":5,\\\"objective\\\":\\\"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\\\",\\\"rationale\\\":\\\"A structured matrix will make the multi-package comparison actionable.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Created artifact-00, a package-by-package matrix covering Python status, grains, texture, KAM/GND, maturity, and recommended role.\\\",\\\"evaluation\\\":\\\"The matrix directly supports package selection and clearly distinguishes full, partial, and non-core capabilities.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"completed\",\n \"result\": \"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\",\n \"evaluation\": \"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a package-by-package matrix covering Python status, grains, texture, KAM/GND, maturity, and recommended role.\",\n \"evaluation\": \"The matrix directly supports package selection and clearly distinguishes full, partial, and non-core capabilities.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_7bcoFGw2YuGzai5ip8f8B7P4", + "info": { + "start_ts": 1015.613473393, + "end_ts": 1015.614209595 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_0Ow8AsXKD75JrGG2nzDUQk6Q", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 6, \"updated_plan_objective_json\": \"{\\\"id\\\":6,\\\"objective\\\":\\\"Call the answer tool with a thorough, evidence-cited technical response.\\\",\\\"rationale\\\":\\\"This is required to submit the final answer after all research objectives are complete.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null}\"}", + "name": "update_plan" + } + } + ], + "info": null + } + ], + "hide_old_env_states": false, + "hide_old_action_content": false, + "sliding_window": null + }, + "next_agent_state": { + "tools": [ + { + "type": "function", + "info": { + "name": "create_plan", + "description": "Create a plan for to construct a thorough and accurate answer.\n\nDecompose the input task into a set of sub-objectives that need to\nbe met in order to form the answer. Subsequently, when working on solving\nthe query, you can focus on these objectives one by one. Note that the\n`update_plan` tool allows you to change the objectives if needed.\n\nIn addition to the objectives you deem necessary, add one final objective\nto call the `answer` tool with your final answer, so that you do not\nforget to submit your answer.", + "parameters": { + "type": "object", + "properties": { + "plan_json": { + "description": "A JSON string representing the plan. It should be a dictionary\nwith top-level key \"objectives\", whose value is a list of objectives.\nEach objective should be a dictionary with the following keys:\n- \"id\": (int) an integer id for the plan objective\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Plan Json", + "type": "string" + } + }, + "required": [ + "plan_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "patent_search", + "description": "Search for patents to increase the patent count.\n\nRepeat previous calls with the same query to continue a search.\nOnly repeat a maximum of three times.\nThis tool can be called concurrently.\nThis tool introduces novel patents, so invoke it when just beginning\nor when unsatisfied with the current evidence.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A search query for patents (using Google Patents). Never include temporal\nfilters here (e.g. specific years or dates); instead use the `after` and `before`\narguments. Never include author filters here; instead use the `inventor` argument.\nSupports the following syntax, but they can be brittle and are not preferred; try\nsimple queries where possible:\n- Boolean operators: AND (default), OR, and - (NOT)\n- Exact phrases: \"quoted text\"\n- Proximity: NEAR/x, ADJ/x (ordered), WITH (20 words), SAME (200 words)\n- Field search: TI=(title), AB=(abstract), CL=(claims)\n- CPC codes: cpc:A01B or CPC=B60R22/low (includes children)\n- Wildcards: ? (0-1 char), * (0+ chars), # (exactly 1 char)\n- Chemistry: exact molecule names, SSS=molecule (substructure),\n ~molecule (similar), SMILES strings, SMARTS=pattern, InChI keys,\n CL=~molecule (chemistry in claims)", + "title": "Query", + "type": "string" + }, + "after": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Minimum patent date. Format: YYYYMMDD.\nExample: '20150101' for patents after Jan 1, 2015.", + "title": "After" + }, + "before": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Maximum patent date. Format: YYYYMMDD.\nExample: '20200101' for patents before Jan 1, 2020.", + "title": "Before" + }, + "date_type": { + "default": "publication", + "description": "Type of date that after and before refer to. Must be either 'publication'\nor 'filing'. Only necessary to set if after and/or before are set.", + "title": "Date Type", + "type": "string" + }, + "inventor": { + "anyOf": [ + { + "type": "string" + }, + { + "type": "null" + } + ], + "default": null, + "description": "Filter by inventor name(s). Comma-separated for multiple.", + "title": "Inventor" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "gather_evidence", + "description": "Given a specific question, gather evidence from full-text of documents that have been retrieved by\nprevious tool calls.\n\nHere is how the tool works: the provided question is posed to all texts retrieved thus far\nby other tools. An LLM sees each text independently and is prompted to find information relevant\nto the question.\n\nThat is, we sample evidence `e \\sim LLM(prompt, text, q)` for each text. We only return pieces of\nevidence `e` that are sufficiently relevant to the question; if the text is irrelevant, then `e`\nis not included in the final response of this tool. Note that this LLM does not see the conversation\nhistory thus far, only the question and each text. Note that it only sees text, no images or other\nmedia from the contents (though tables and formulae may be parseable in some cases).\n\nThink of this tool as an efficient way to deeply scan all retrieved documents for evidence. It\nis complementary to the `read_text` tool, which can be used to directly inspect a subset of the\nretrieved documents.\n\nA valuable time to invoke this tool is after another tool finds potentially relevant documents.\nFeel free to invoke this tool in parallel with other tools, but do not call this tool in parallel with itself.\nOnly invoke this tool when the document count is above zero, or this tool will be useless.\n\nIf you call this tool 2+ times and repeatedly get 'No relevant evidence found',\nconsider whether the answer might be in visual content (figures/tables) rather than text,\nand use appropriate tools if available to access that content.\n\nOnce you receive evidence from this tool:\n\n1. Read Carefully: Don't skim. The evidence may contain multiple similar terms\n or values\u2014make sure you identify the RIGHT one.\n\n2. For Comparative Questions: If the question asks \"which X has greatest Y\",\n check that the evidence includes information about ALL relevant X options, not\n just one or two. If you only see partial information, gather more evidence.\n\n3. Check for Noise: Verify that numbers or terms aren't just metadata\n (page numbers, reference IDs, dates). Look for the value in meaningful\n scientific context.\n\n4. Context Matters: A term appearing 71 times doesn't mean it's always the answer.\n Verify which occurrence actually answers the question asked.\n\nIf the evidence doesn't clearly answer the question, gather more evidence before\nconcluding. Don't guess based on partial information.", + "parameters": { + "type": "object", + "properties": { + "question": { + "description": "Specific question to gather evidence for.", + "title": "Question", + "type": "string" + }, + "text_name_focus_list": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "if there are known text names that are relevant to the question,\nyou can specify them here. This will ensure that the tool will look at these text names\nfirst, and then look at the rest of the evidence. If left as `None`, an embedding\nsearch will be performed on underlying content to find most relevant texts.", + "title": "Text Name Focus List" + } + }, + "required": [ + "question" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "update_plan", + "description": "Update the plan for a specific or new objective using a JSON string to create or\nreplace the existing objective.\n\nThe updated_json should be a JSON string representing a single plan objective object.\n\nYou may use this tool to update an objective, create a new one, or change\nstrategy if the current plan is not working. Only change strategy significantly\nas a last resort, after trying very hard to make the current plan work.\n\nOnly mark an objective as completed if you are absolutely sure that it has been fully\nmet. If it is only partially met, you may add additional objectives to the plan for\nwhat is missing.", + "parameters": { + "type": "object", + "properties": { + "plan_objective_id": { + "description": "The id of the objective to update.", + "title": "Plan Objective Id", + "type": "integer" + }, + "updated_plan_objective_json": { + "description": "A JSON string representing the updated objective.\nIt should be a dictionary with the with the following keys:\n- \"id\": (int) an integer id matching plan_objective_id, if not found, a new objective is created.\n- \"objective\": (str) a human readable description of the objective's goal\n- \"rationale\": (str) a human readable rationale\n describing why this objective is necessary and appropriate for the query.\n- \"status\": (str) updated task status, default to pending,\n one of: pending, in-progress, or completed.\n \"pending\" is yet to be started,\n \"in-progress\" is the current objective being solved,\n \"completed\" is successfully completed with acceptance criteria met.\n- \"result\": (str|None) The result of the objective, default to None.\n- \"evaluation\": (str|None) Description of why the objective result\n meets the criteria for being complete. Default to None.", + "title": "Updated Plan Objective Json", + "type": "string" + } + }, + "required": [ + "plan_objective_id", + "updated_plan_objective_json" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "paper_search", + "description": "Search for papers in Google Scholar for you to look at.\n\nCan be called repeatedly with identical parameters to retrieve more results (max 2 repeats).\nA short summary of the retrieved papers will be returned, but to examine the full content of each paper,\nyou must use the gather_evidence tool.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "A Google Scholar search query, which can be a specific phrase, complete sentence,\nor general keywords, e.g. 'machine learning for immunology'. Also can be\ngiven search operators like 'author:'.", + "title": "Query", + "type": "string" + }, + "min_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for minimum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Min Year" + }, + "max_year": { + "anyOf": [ + { + "type": "integer" + }, + { + "type": "null" + } + ], + "default": null, + "description": "filter for maximum publication year.\nThe current year is 2026. If unsure about time range, do not set.", + "title": "Max Year" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "clinical_trials_search", + "description": "Search for clinical trials, with support for repeated calls and concurrent execution.\n\nWill add new clinical trials to the state, and return metadata about the number of trials found.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "The search query string. Supports complex boolean expressions, field-specific\nsearches, and query modifiers through operators. All configuration is done through\noperators in the query string.\nQuery Syntax:\n Basic Search:\n Simple text automatically uses default EXPANSION[Relaxation] and COVERAGE[Contains]\n >>> \"heart attack\"\n\n Modified Search:\n Use operators to modify search behavior:\n >>> 'EXPANSION[None]COVERAGE[FullMatch]\"exact phrase\"'\n >>> 'EXPANSION[Concept]heart attack'\n\n Field Search:\n Specify fields using AREA operator:\n >>> 'AREA[InterventionName]aspirin'\n >>> 'AREA[Phase]PHASE3'\n\n Location Search:\n Use SEARCH operator for compound location queries:\n >>> 'cancer AND SEARCH[Location](AREA[LocationCity]Boston AND AREA[LocationState]Massachusetts)'\n\n Complex Boolean:\n Combine terms with AND, OR, NOT and parentheses:\n >>> '(cancer OR tumor) AND NOT (EXPANSION[None]pediatric OR AREA[StdAge]CHILD)'\n\n Date Ranges:\n Use RANGE to specify date ranges with formats like \"yyyy-MM\" or \"yyyy-MM-dd\".\n Note that MIN and MAX can be used for open-ended ranges:\n >>> AREA[ResultsFirstPostDate]RANGE[2015-01-01, MAX]\n\nOperators:\n EXPANSION[type]: Controls term expansion\n - None: Exact match only, case and accent sensitive\n - Term: Includes lexical variants (plurals, spellings)\n - Concept: Includes UMLS synonyms\n - Relaxation: Relaxes adjacency requirements (default)\n - Lossy: Allows missing partial terms\n\n COVERAGE[type]: Controls text matching\n - FullMatch: Must match entire field\n - StartsWith: Must match beginning of field\n - EndsWith: Must match end of field\n - Contains: Must match part of field (default)\n\n AREA[field]: Specifies field to search\n - See Field Reference for available fields\n\n SEARCH[type]: Groups field searches\n - Location: Groups location-related fields\n - Study: Groups study-related fields\n\nUsage Notes:\n - All search expressions are implicitly OR expressions\n - Operator precedence (highest to lowest): terms/source operators, NOT/context operators, AND, OR\n - Use quotes for exact phrase matching: \"heart attack\"\n - Use parentheses for grouping: (heart OR cardiac) AND attack\n - Use backslash to escape operators: \\AND\n - Default expansion is EXPANSION[Relaxation]\n - Default coverage is COVERAGE[Contains]\n\nField Reference:\n High Priority Fields (weight >= 0.8):\n - NCTId (1.0): Trial identifier\n - Acronym (1.0): Study acronym\n - BriefTitle (0.89): Short title\n - OfficialTitle (0.85): Full official title\n - Condition (0.81): Medical condition\n - InterventionName (0.8): Primary intervention name\n - OverallStatus: Trial status\n\n Medium Priority Fields (0.5-0.79):\n - InterventionOtherName (0.75): Alternative intervention names\n - Phase (0.65): Trial phase\n - StdAge (0.65): Standard age groups\n - Keyword (0.6): Study keywords\n - BriefSummary (0.6): Short description\n - SecondaryOutcomeMeasure (0.5): Secondary outcomes\n\n Low Priority Fields (< 0.5):\n - DesignPrimaryPurpose (0.3): Primary purpose of study\n - StudyType (0.3)\n - Various descriptive, location, and administrative fields\n\nSupported Enums:\n Phase:\n - EARLY_PHASE1: Early Phase 1\n - PHASE1: Phase 1\n - PHASE2: Phase 2\n - PHASE3: Phase 3\n - PHASE4: Phase 4\n - NA: Not Applicable\n\n StandardAge:\n - CHILD: Child\n - ADULT: Adult\n - OLDER_ADULT: Older Adult\n\n Status:\n - RECRUITING: Currently recruiting participants\n - ACTIVE_NOT_RECRUITING: Active but not recruiting\n - COMPLETED: Study completed\n - ENROLLING_BY_INVITATION: Enrolling by invitation only\n - NOT_YET_RECRUITING: Not yet recruiting\n - SUSPENDED: Study suspended\n - TERMINATED: Study terminated\n - WITHDRAWN: Study withdrawn\n - AVAILABLE: Available\n - NO_LONGER_AVAILABLE: No longer available\n - TEMPORARILY_NOT_AVAILABLE: Temporarily not available\n - APPROVED_FOR_MARKETING: Approved for marketing\n - WITHHELD: Withheld\n - UNKNOWN: Unknown status\n\n StudyType:\n - INTERVENTIONAL: Interventional studies\n - OBSERVATIONAL: Observational studies\n - EXPANDED_ACCESS: Expanded access studies\n\n PrimaryPurpose:\n - TREATMENT: Treatment\n - PREVENTION: Prevention\n - DIAGNOSTIC: Diagnostic\n - ECT: Educational/Counseling/Training\n - SUPPORTIVE_CARE: Supportive Care\n - SCREENING: Screening\n - HEALTH_SERVICES_RESEARCH: Health Services Research\n - BASIC_SCIENCE: Basic Science\n - DEVICE_FEASIBILITY: Device Feasibility\n - OTHER: Other\n\n InterventionType:\n - BEHAVIORAL: Behavioral interventions\n - BIOLOGICAL: Biological interventions\n - COMBINATION_PRODUCT: Combination product interventions\n - DEVICE: Device interventions\n - DIAGNOSTIC_TEST: Diagnostic test interventions\n - DIETARY_SUPPLEMENT: Dietary supplement interventions\n - DRUG: Drug interventions\n - GENETIC: Genetic interventions\n - PROCEDURE: Procedure interventions\n - RADIATION: Radiation interventions\n - OTHER: Other interventions\n\n DesignAllocation:\n - RANDOMIZED: Randomized allocation\n - NON_RANDOMIZED: Non-randomized allocation\n - NA: Not applicable\n\n InterventionalAssignment:\n - SINGLE_GROUP: Single group assignment\n - PARALLEL: Parallel assignment\n - CROSSOVER: Crossover assignment\n - FACTORIAL: Factorial assignment\n - SEQUENTIAL: Sequential assignment\n\n ObservationalModel:\n - COHORT: Cohort\n - CASE_CONTROL: Case-Control\n - CASE_ONLY: Case-Only\n - CASE_CROSSOVER: Case-Crossover\n - ECOLOGIC_OR_COMMUNITY: Ecologic or Community\n - FAMILY_BASED: Family-Based\n - DEFINED_POPULATION: Defined Population\n - NATURAL_HISTORY: Natural History\n - OTHER: Other\n\n DesignMasking:\n - NONE: None (Open Label)\n - SINGLE: Single\n - DOUBLE: Double\n - TRIPLE: Triple\n - QUADRUPLE: Quadruple\n\n WhoMasked:\n - PARTICIPANT: Participant\n - CARE_PROVIDER: Care Provider\n - INVESTIGATOR: Investigator\n - OUTCOMES_ASSESSOR: Outcomes Assessor", + "title": "Query", + "type": "string" + } + }, + "required": [ + "query" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "view_images", + "description": "Retrieve images from a document.\n\nUse this tool when you need to see figures, tables, or other visual content.\nThe tool works by asking a weaker LLM agent to scan through the document, looking for\nvisual content matching the query you provide. When the agent finds matches, it will\ncrop the page to the relevant region. It can be tasked to return multiple images,\nif desired. Note that the weaker LLM may be worse at *understanding* images than\nyou, so double-check its conclusions about the returned images. It is also stateless,\ni.e. it will not remember what you have asked it to find in the past.\n\nUse this tool carefully. It is difficult to retrieve images, and the images\nthemselves will fill up your context window quickly. Here are some guidelines:\n1. ALWAYS use this tool if the question explicitly mentions 'Table X' or 'Figure Y'.\n These questions require visual content and cannot be answered with text alone.\n2. Use this tool as a fallback when gather_evidence fails repeatedly (2+ times\n returning 'No relevant evidence'). This suggests the answer may be in visual content.\n3. When gather_evidence returns text mentioning 'see Figure X' or 'shown in Table Y',\n you should call this tool to examine that specific visual reference.\n4. Only call this tool after directly inspecting relevant text with read_text\n to confirm the document likely contains the visual you need.\n5. Don't use this tool in a broad-spectrum way (don't call on many documents\n just to see what might be relevant).\n6. Never call this tool in parallel with itself.\n7. Users like visual evidence, so call this tool at least once before answering so you\n can provide some image citations in the answer. Add this to your plan so you do not forget.\n For images to be presented to the user, they must be cited in the final answer after retrieval\n by this tool. You may ignore this instruction if the request is for the question is asking\n for a specific piece of information and is not an open-ended research question.\n8. NEVER cite full-page images in your final answer, unless they are full-page panels figures or\n tables. Cite relevant cropped regions if available, otherwise consider calling this tool again\n to find a smaller region that answers the query.", + "parameters": { + "type": "object", + "properties": { + "query": { + "description": "Query to search for visual content in the document. If you know the specific\nfigure/table/formula you are looking for, you can specify it here (e.g. \"find me\nFigure 3\" or \"extract the table on page 10\"). You can also request based on content\n(e.g. \"find me all volcano plots of gene X\" or \"I need the formula for the growth rate\").\nYou can even include explicit instructions for the LLM agent to follow (if helpful, not\nnecessarily required). The more specific and clear your query is here, the higher the\nchance of success.", + "title": "Query", + "type": "string" + }, + "text_name": { + "description": "Name of a text chunk (e.g., 'smith2023example pages 1-5').", + "title": "Text Name", + "type": "string" + } + }, + "required": [ + "query", + "text_name" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "read_text", + "description": "Given a list of text names, read their contents.\n\nThis tool is valuable to read the full contents of specific parts of documents,\ncomplementing the `gather_evidence` tool's indirect approach.\nThis can be an especially useful tool if `gather_evidence` does not find evidence,\nbut you think it is very likely that specific documents are still relevant to the question.", + "parameters": { + "type": "object", + "properties": { + "text_name_list": { + "description": "A list of up to 10 text names to read.\nThese text names can be any valid text names retrieved by other tools.", + "items": { + "type": "string" + }, + "title": "Text Name List", + "type": "array" + } + }, + "required": [ + "text_name_list" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "create_artifact", + "description": "Create an artifact that can be used in the final answer, based on information\nyou have collected so far.\n\nHow it works: the full message history thus far is passed to another LLM,\nalong with the question and your description. The LLM is tasked to create\nan artifact that matches your specifications, including a title, caption,\nand markdown content. You drive the process by specifying the type and\ndescribing what you want.\n\nOnly use this tool *after* you have gathered sufficient evidence or\ninformation to create a new artifact. Do not call this tool in parallel\nwith tools that are searching for information or evidence that would be\nnecessary to create this artifact.\n\nIt is very helpful after calls to several search tools,\nafter calls to the `gather_evidence` tool, but it is not required.\nHowever, it should be called before the `answer` tool at\nleast one time, but more may be given if specifically helpful for\nthe task, so that the `answer` tool can use the artifacts for the answer.\n\nAn artifact can be any string representable fact or list of facts,\nformatted using markdown syntax. The artifact will be generated as one of\nthe following types:\n- table: Markdown tables with rows and columns (e.g., findings, methods, molecules)\n- code_block: Code snippets in fenced code blocks\n- blockquote: Quoted text or important facts (using > prefix)\n- image: AI-generated image (chart, diagram, or visualization)\n\nEach artifact consists of:\n- A short title (3-8 words) that names what the artifact contains\n- A caption (1-2 sentences) describing what the artifact shows and why it's useful\n- Markdown content in a single element type (validated automatically with retries if needed)\n For image type, content is a markdown image embed ``![title](data:image/png;base64,...)``,\n embedding the PNG directly as a base64 data URL.", + "parameters": { + "type": "object", + "properties": { + "artifact_type": { + "description": "Artifact type to create. Must be one of: table,\ncode_block, blockquote, image.", + "title": "Artifact Type", + "type": "string" + }, + "description": { + "description": "Description of what you want in the artifact. Be specific\nabout what information should be included, how it should be organized,\nand what columns/sections/content you want. This guides the summary LLM.\nFor image type, describe the visual you want generated in detail.", + "title": "Description", + "type": "string" + } + }, + "required": [ + "artifact_type", + "description" + ] + } + } + }, + { + "type": "function", + "info": { + "name": "open_targets_search", + "description": "Search for disease-target associations from OpenTargets database.\n\nAdds new disease-target association data to the state and returns top search results.", + "parameters": { + "type": "object", + "properties": { + "disease_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of disease names to search for.\nUsage:\ndisease_names = [\"Alzheimer's\"] ->\n Returns top targets for Alzheimer's\ndisease_names = [\"Alzheimer's\", \"Diabetes\"], target_names = [\"PSEN1\"] ->\n Filters for only PSEN1-Alzheimer's/Diabetes evidence\ndisease_names = [\"Alzheimer's\"], drug_names = [\"Aducanumab\"] ->\n Filters for Alzheimer's disease-target pairs treated by Aducanumab", + "title": "Disease Names" + }, + "target_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of target names to search for.\nUsage:\ntarget_names = [\"BACE1\"] ->\n Returns top diseases associated with BACE1\ntarget_names = [\"BACE1\", \"APP\"], disease_names = [\"Alzheimer's\"] ->\n Filters for only these target-Alzheimer's pairs\ntarget_names = [\"BACE1\"], drug_names = [\"Aducanumab\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Target Names" + }, + "drug_names": { + "anyOf": [ + { + "items": { + "type": "string" + }, + "type": "array" + }, + { + "type": "null" + } + ], + "default": null, + "description": "List of drug names to generate or filter results.\nUsage:\ndrug_names = [\"Aducanumab\"] ->\n Returns all disease-target pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], disease_names = [\"Alzheimer's\"] ->\n Filters for Alzheimer's pairs treated by Aducanumab\ndrug_names = [\"Aducanumab\"], target_names = [\"BACE1\"] ->\n Filters for BACE1 disease pairs treated by Aducanumab", + "title": "Drug Names" + } + }, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "collect_cited_papers_in_evidence", + "description": "Collect papers by traversing the citations of relevant papers to increase the paper count.\n\nThis tool has no effect if called when paper count or relevant papers are zero.\nThis tool will find papers that are likely to lead to more relevant evidence.\nThe papers found have a higher chance of relevance because this tool starts from relevant evidence.\nThis tool's usefulness increases with the amount of gathered evidence as there's more citations to consider.", + "parameters": { + "type": "object", + "properties": {}, + "required": [] + } + } + }, + { + "type": "function", + "info": { + "name": "answer", + "description": "Submit your final answer to the user. You MUST call this tool on your final answer.\nIf you mark all objectives as completed but have not yet called this tool, then\nthe answer will not be submitted. Note that this tool should only be called once, as it\nimmediately terminates the run and submits the answer.\n\nWrite an answer based on the evidence from gather_evidence results in message history.\nIf the context provides insufficient information, provide a partial answer and explain\nwhy it is partial (e.g. the question is currently an open question in the field,\nyou were unable to obtain certain papers, etc.).\n\nFor each part of your answer, indicate which sources most support it via citation IDs at\nthe end of sentences, like (pqac-0000ABCD).\nYou are only allowed to cite available context IDs, indicated in previous messages by \"Context IDs that may be cited:\".\n## Valid citation examples, only use comma/space delimited parentheticals:\n\n- (pqac-0000ABCD, pqac-0000EFGH)\n- (pqac-0000IJKL)\n\n## Invalid citation examples:\n\n- (pqac-0000ABCD and pqac-0000EFGH)\n- (pqac-0000ABCD;pqac-0000EFGH)\n- (pqac-0000ABCD-pqac-0000EFGH)\n- pqac-0000ABCD and pqac-0000EFGH\n- Example's work pqac-0000ABCD\n- (pages pqac-0000ABCD)\n- ![description](pqac-0000ABCD)\n- ()\n\nDo not concatenate citation ids, just use them as-is.\n\nWhen writing your answer, embed artifacts if relevant artifacts have been previously\ncreated by the `create_artifact` tool. To embed an artifact, simply include this in your\nanswer, on its own line and without any additional formatting (i.e. no backticks, code fences,\nquotes, additional tags, etc.).\nAlways provide at least some introductory sentence or paragraph before referring to an artifact,\nso that the artifact is properly contextualized.\nHere is an example reference to an artifact with ID \"artifact-XY\":\n\n\nValid artifact IDs are indicated in previous messages by 'Artifact IDs that may be injected into the answer:'.\n\n\nDouble-check your answer:\n\n1. Evidence Match: Re-read the specific evidence snippets you're citing. Does your\n answer actually match what they say? Don't rely on memory from 20+ messages ago.\n\n2. For Comparative Questions (\"which X has greatest/maximum/highest Y\"):\n - Did you examine ALL options mentioned in the evidence?\n - Did you explicitly compare their values?\n - Example: If evidence mentions \"Pancreatic: 0.92, Liver: 0.85, Lung: 0.78\",\n verify you compared all three, not just two.\n\n3. For Specific Value Questions (\"How many\", \"What is the\", specific terms):\n - Does the exact value you're stating appear in the evidence?\n - Are you extracting from the right context (not noise like page numbers)?\n\n4. Sanity Check: Read your answer. Does it directly address what was asked?\n\nDon't use Markdown image syntax like ![...](...) to reference media in your answer,\nbecause media get automatically linked by the system based on your answer's citations.\n\nWrite in the style of a scientific article, with concise sentences and\ncoherent paragraphs. This answer will be used directly,\nso do not add any extraneous information.", + "parameters": { + "type": "object", + "properties": { + "answer": { + "description": "valid answer, cited with citation IDs.", + "title": "Answer", + "type": "string" + } + }, + "required": [ + "answer" + ] + } + } + } + ], + "messages": [ + { + "role": "user", + "content": "Use the tools to answer the question: What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.\n\nStart by making a plan, and update the plan as you go. Your answer will be returned to the user when all plan objectives are completed. Be thorough and ensure your evidence matches the user's question.", + "info": null + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_kcmzfiVWrTdIWQyBi42VuIon", + "type": "function", + "function": { + "arguments": "{\"plan_json\": \"{\\\"objectives\\\":[{\\\"id\\\":1,\\\"objective\\\":\\\"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\\\",\\\"rationale\\\":\\\"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":2,\\\"objective\\\":\\\"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\\\",\\\"rationale\\\":\\\"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":3,\\\"objective\\\":\\\"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\\\",\\\"rationale\\\":\\\"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":4,\\\"objective\\\":\\\"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\\\",\\\"rationale\\\":\\\"The answer should support a concrete implementation decision, not merely list software.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":5,\\\"objective\\\":\\\"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\\\",\\\"rationale\\\":\\\"A structured matrix will make the multi-package comparison actionable.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null},{\\\"id\\\":6,\\\"objective\\\":\\\"Call the answer tool with a thorough, evidence-cited technical response.\\\",\\\"rationale\\\":\\\"This is required to submit the final answer after all research objectives are complete.\\\",\\\"status\\\":\\\"pending\\\",\\\"result\\\":null,\\\"evaluation\\\":null}]}\"}", + "name": "create_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Plan registered and formatted correctly, with some modifications:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "create_plan", + "tool_call_id": "call_kcmzfiVWrTdIWQyBi42VuIon", + "info": { + "start_ts": 615.740379951, + "end_ts": 615.744231214 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_VIQzh60P2I1sB79k1x1yJiua", + "type": "function", + "function": { + "arguments": "{\"query\": \"kikuchipy open source Python EBSD software paper orix orientation mapping grain segmentation KAM\", \"min_year\": 2018}", + "name": "paper_search" + } + }, + { + "id": "call_26RNCIR7p5ooimL1oHJKDxPe", + "type": "function", + "function": { + "arguments": "{\"query\": \"DefDAP Python EBSD grain reconstruction texture KAM GND software paper\", \"min_year\": 2018}", + "name": "paper_search" + } + }, + { + "id": "call_TWgXFKMItXYFpWEbgg83FfU9", + "type": "function", + "function": { + "arguments": "{\"query\": \"DREAM.3D software grain segmentation EBSD texture analysis paper simplnx Python\", \"min_year\": 2010}", + "name": "paper_search" + } + }, + { + "id": "call_oB8LA7Kpebnw5E9YNkkGkAve", + "type": "function", + "function": { + "arguments": "{\"query\": \"pyxem orientation mapping EBSD grain segmentation texture analysis software paper\", \"min_year\": 2016}", + "name": "paper_search" + } + }, + { + "id": "call_sXDuxIhQcjT0pGraCjqsLz2r", + "type": "function", + "function": { + "arguments": "{\"query\": \"MTEX texture ODF grain reconstruction KAM GND density EBSD software\", \"min_year\": 2005}", + "name": "paper_search" + } + }, + { + "id": "call_I26KQDTI4oHtsQAWeLYGK9RU", + "type": "function", + "function": { + "arguments": "{\"query\": \"Python EBSD post processing grain reconstruction texture pole figure ODF KAM GND open source\"}", + "name": "paper_search" + } + }, + { + "id": "call_5NDupt5Es4nrgT9SSGt0veGH", + "type": "function", + "function": { + "arguments": "{\"query\": \"pyEBSD open source Python EBSD grain KAM GND\"}", + "name": "paper_search" + } + }, + { + "id": "call_TcPaoKE5klpaxx5d8PlBUDrk", + "type": "function", + "function": { + "arguments": "{\"query\": \"pymicro crystallographic texture EBSD grain Python software\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Machine learning enhanced analysis of EBSD data for texture representation\nValid Text Names:\n'wanni2024machinelearningenhanced pages 1-2', 'wanni2024machinelearningenhanced pages 2-3', 'wanni2024machinelearningenhanced pages 3-5', 'wanni2024machinelearningenhanced pages 5-6', 'wanni2024machinelearningenhanced pages 6-7', 'wanni2024machinelearningenhanced pages 7-9', 'wanni2024machinelearningenhanced pages 9-10', 'wanni2024machinelearningenhanced pages 10-11'\n\nBibTex:\n@article{wanni2024machinelearningenhanced,\n author = \"Wanni, J. and Bronkhorst, C. A. and Thoma, D. J.\",\n title = \"Machine learning enhanced analysis of EBSD data for texture representation\",\n year = \"2024\",\n journal = \"npj Computational Materials\",\n volume = \"10\",\n pages = \"1-11\",\n month = \"Jun\",\n doi = \"10.1038/s41524-024-01324-4\",\n url = \"https://doi.org/10.1038/s41524-024-01324-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n\n\nAbstract:\nGenerating reduced-order, synthetic grain structure datasets that accurately represent the measured grain structure of a material is important for reducing the cost and increasing the accuracy of computational crystal plasticity efforts. This study introduces a machine-learning-based approach, termed texture adaptive clustering and sampling (TACS), for generating representative Euler angle datasets that accurately mimic the crystallographic texture. The TACS approach employs K-means clustering and density-based sampling in a closed-loop iteration to create representative Euler angle datasets. Proof-of-principle experiments were performed on rolled and recrystallized low-carbon steel. Validation of the TACS approach was extended to twenty-two datasets, varying lattice structures, and complex crystallographic textures, thereby encompassing a broad range of materials and crystal structures. Kolmogorov-Smirnov (K-S) test comparisons underscore the performance of the TACS approach over traditional electron backscatter diffraction EBSD dataset reduction techniques, with average K-S test scores nearing 0.9, indicating a high-fidelity representation of the original datasets. In contrast, conventional methods display scores below 0.3, indicating less reliability of the structure representation. The independence of the TACS approach from material texture and its capability to autonomously generate datasets with predetermined data points demonstrates its unbiased potential in streamlining dataset preparation for crystallographic analysis.\n\nRelevant Snippet:\n\u2026 and kernel average misorientation (KAM) maps to deconvolute \u2026 of MTEX, a MATLAB toolbox for texture analysis, the ODF is \u2026 kernel density estimation (KDE), kernel density estimation \u2026\n\n\n---\n\n## 2. Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy\nValid Text Names:\n'bachmann2022efficientreconstructionof pages 1-2', 'bachmann2022efficientreconstructionof pages 2-3', 'bachmann2022efficientreconstructionof pages 3-5', 'bachmann2022efficientreconstructionof pages 5-6', 'bachmann2022efficientreconstructionof pages 6-7', 'bachmann2022efficientreconstructionof pages 7-10', 'bachmann2022efficientreconstructionof pages 10-11', 'bachmann2022efficientreconstructionof pages 11-15', 'bachmann2022efficientreconstructionof pages 15-16', 'bachmann2022efficientreconstructionof pages 16-17', 'bachmann2022efficientreconstructionof pages 17-18', 'bachmann2022efficientreconstructionof pages 18-18'\n\nBibTex:\n@article{bachmann2022efficientreconstructionof,\n author = \"Bachmann, Bj\u00f6rn-Ivo and M\u00fcller, Martin and Britz, Dominik and Durmaz, Ali Riza and Ackermann, Marc and Shchyglo, Oleg and Staudt, Thorsten and M\u00fccklich, Frank\",\n title = \"Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy\",\n year = \"2022\",\n journal = \"Frontiers in Materials\",\n volume = \"9\",\n month = \"Oct\",\n doi = \"10.3389/fmats.2022.1033505\",\n url = \"https://doi.org/10.3389/fmats.2022.1033505\",\n publisher = \"Frontiers Media SA\",\n issn = \"2296-8016\"\n}\n\n\nAbstract:\nThe high-temperature austenite phase is the initial state of practically all technologically relevant hot forming and heat treatment operations in steel processing. The phenomena occurring in austenite, such as recrystallization or grain growth, can have a decisive influence on the subsequent properties of the material. After the hot forming or heat treatment process, however, the austenite transforms into other microstructural constituents and information on the prior austenite morphology are no longer directly accessible. There are established methods available for reconstructing former austenite grain boundaries via metallographic etching or electron backscatter diffraction (EBSD) which both exhibit shortcomings. While etching is often difficult to reproduce and strongly depend on the investigated steel\u2019s alloying concept, EBSD acquisition and reconstruction is rather time-consuming. But in fact, though, light optical micrographs of steels contrasted with conventional Nital etchant also contain information about the former austenite grains. However, relevant features are not directly apparent or accessible with conventional segmentation approaches. This work presents a deep learning (DL) segmentation of prior austenite grains (PAG) from Nital etched light optical micrographs. The basis for successful segmentation is a correlative characterization from EBSD, light and scanning electron microscopy to specify the ground truth required for supervised learning. The DL model shows good and robust segmentation results. While the intersection over union of 70% does not fully reflect the model performance due to the inherent uncertainty in PAG estimation, a mean error of 6.1% in mean grain size derived from the segmentation clearly shows the high quality of the result.\n\nRelevant Snippet:\n\u2026 segmentation is a correlative characterization from EBSD, \u2026 orientation data from electron backscatter diffraction (EBSD) \u2026 This paper represents the first publication of the consortiums\u2019 \u2026\n\n\nUnobtainable Papers:\nDavid J. Rowenhorst, Patrick G. Callahan, and H\u00e5kon W. \u00c5nes. Fast radon transforms for high-precision ebsd orientation determination using pyebsdindex. Journal of Applied Crystallography, 57:3-19, Feb 2024. URL: https://doi.org/10.1107/s1600576723010221, doi:10.1107/s1600576723010221.\nS. Breumier, T. Martinez Ostormujof, B. Frincu, N. Gey, A. Couturier, N. Loukachenko, P.E. Aba-perea, and L. Germain. Leveraging ebsd data by deep learning for bainite, ferrite and martensite segmentation. Materials Characterization, 186:111805, Apr 2022. URL: https://doi.org/10.1016/j.matchar.2022.111805, doi:10.1016/j.matchar.2022.111805.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_VIQzh60P2I1sB79k1x1yJiua", + "info": { + "start_ts": 625.522755479, + "end_ts": 663.10598677 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nIoannis Alakiozidis, Callum Hunt, Albert D. Smith, Mia Maric, Zaheen Shah, A. Ambard, and Philipp Frankel. Microstructure and mechanical performance of cold spray cr coatings. Nov 2024. URL: https://doi.org/10.5281/zenodo.14034890, doi:10.5281/zenodo.14034890.\nPrzemys\u0142aw Snopi\u0144ski and Ond\u0159ej Hil\u0161er. Mechanism of grain refinement in 3d-printed alsi10mg alloy subjected to severe plastic deformation. Materials, 17:4098, Aug 2024. URL: https://doi.org/10.3390/ma17164098, doi:10.3390/ma17164098.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_26RNCIR7p5ooimL1oHJKDxPe", + "info": { + "start_ts": 625.558711009, + "end_ts": 664.725048365 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach\nValid Text Names:\n'yi2025rapidgrainsegmentation pages 1-2', 'yi2025rapidgrainsegmentation pages 2-4', 'yi2025rapidgrainsegmentation pages 4-7', 'yi2025rapidgrainsegmentation pages 7-9', 'yi2025rapidgrainsegmentation pages 9-11', 'yi2025rapidgrainsegmentation pages 11-13', 'yi2025rapidgrainsegmentation pages 13-14'\n\nBibTex:\n@article{yi2025rapidgrainsegmentation,\n author = \"Yi, Yu-Tsen and Seo, Junwon and Murphy, Kevin and Rollett, Anthony D.\",\n title = \"Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach\",\n year = \"2025\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jan\",\n doi = \"10.1007/s40192-024-00390-2\",\n url = \"https://doi.org/10.1007/s40192-024-00390-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\n\n Grain size distribution is a critical factor in determining materials\u2019 physical and mechanical properties, including thermal conductivity, hardness, and creep behavior. Understanding the distribution of grain sizes is essential for advancing the comprehension of material properties and improving materials development and design. Traditional methods for determining grain size, such as electron backscatter diffraction (EBSD), are resource-intensive, underscoring the need for more efficient approaches to grain segmentation in standard micrographs, such as those obtained via SEM and optical imaging. This paper presents a streamlined, unsupervised computer vision pipeline that employs superpixel segmentation and region adjacency merging techniques to segment and measure grain geometry from micrographs efficiently. The pipeline is validated using two methods: hand-labeled SEM images of laser powder bed fusion (LPBF) fabricated Haynes 282 Ni-alloy and open-source EBSD data of IN100 from Dream3D. Both validation approaches achieved IoU and Dice scores greater than 0.9, while processing an image with a resolution of 1000 \u00d7 1000 pixels in under 40\u00a0s, demonstrating a fast and sufficiently accurate pipeline.\n\nRelevant Snippet:\n\u2026 (EBSD) data and Dream3D software to generate grain boundary maps for comparison with the segmentation \u2026 Uchic et al. from the Air Force Research Laboratory (AFRL), was used in \u2026\n\n\n---\n\n## 2. Modeling complex polycrystalline alloys using a Generative Adversarial Network enabled computational platform\nValid Text Names:\n'murgas2024modelingcomplexpolycrystalline pages 1-2', 'murgas2024modelingcomplexpolycrystalline pages 2-3', 'murgas2024modelingcomplexpolycrystalline pages 3-4', 'murgas2024modelingcomplexpolycrystalline pages 4-6', 'murgas2024modelingcomplexpolycrystalline pages 6-7', 'murgas2024modelingcomplexpolycrystalline pages 7-10', 'murgas2024modelingcomplexpolycrystalline pages 10-11', 'murgas2024modelingcomplexpolycrystalline pages 11-12', 'murgas2024modelingcomplexpolycrystalline pages 12-13', 'murgas2024modelingcomplexpolycrystalline pages 13-14', 'murgas2024modelingcomplexpolycrystalline pages 14-15', 'murgas2024modelingcomplexpolycrystalline pages 15-16', 'murgas2024modelingcomplexpolycrystalline pages 16-16'\n\nBibTex:\n@article{murgas2024modelingcomplexpolycrystalline,\n author = \"Murgas, Brayan and Stickel, Joshua and Brewer, Luke and Ghosh, Somnath\",\n title = \"Modeling complex polycrystalline alloys using a Generative Adversarial Network enabled computational platform\",\n year = \"2024\",\n journal = \"Nature Communications\",\n volume = \"15\",\n month = \"Nov\",\n doi = \"10.1038/s41467-024-53865-3\",\n url = \"https://doi.org/10.1038/s41467-024-53865-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2041-1723\"\n}\n\n\nAbstract:\nCreating statistically equivalent virtual microstructures (SEVM) for polycrystalline materials with complex microstructures that encompass multi-modal morphological and crystallographic distributions is a challenging enterprise. Cold spray-formed (CSF) AA7050 alloy containing coarse-grained prior particles and ultra-fine grains (UFG) and additively manufactured (AM) Ti64 alloys with alpha laths in beta substrates. The paper introduces an approach strategically integrating a Generative Adversarial Network (GAN) for multi-modal microstructures with a synthetic microstructure builder DREAM.3D for packing grains conforming to statistics in electron backscatter diffraction (EBSD) maps for generating SEVMs of CSF and AM alloy microstructures. A robust multiscale model is subsequently developed for self-consistent coupling of crystal plasticity finite element model (CPFEM) for coarse-grained crystals with an upscaled constitutive model for UFGs. Sub-volume elements are simulated for efficient computations and their responses are averaged for overall stress-strain response. The methods developed are important for image-based micromechanical modeling that is necessary for microstructure-property relations.\n\nRelevant Snippet:\n\u2026 12e, f, the SEVM values lie between those of the EBSD maps. The crystallographic orientation of every variant is taken from the EBSD data, and the grains are segmented. Figure 5d\u2013f \u2026\n\n\n---\n\n## 3. Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\nValid Text Names:\n'stanek2020comparisonofsegmentation pages 1-2', 'stanek2020comparisonofsegmentation pages 2-7', 'stanek2020comparisonofsegmentation pages 7-9', 'stanek2020comparisonofsegmentation pages 9-12', 'stanek2020comparisonofsegmentation pages 12-16', 'stanek2020comparisonofsegmentation pages 16-18', 'stanek2020comparisonofsegmentation pages 18-19'\n\nBibTex:\n@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n\n\nAbstract:\nA comparison study of grain segmentation in 2D and 3D electron backscatter diffraction (EBSD) measurements of polycrystalline materials, based on a misorientation threshold, is presented. Using Laguerre tessellations with orientation marks, a simulation study that demonstrates a different sensitivity of segmentation in the case of low angle and high angle grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investigated concerning segmentation, and there are given some recommendations for the misorientation thresholds. Computational tools such as DREAM 3D software enable significant data evaluation that is completed by a five-parameter analysis of grain boundaries. K e y w o r d s: electron backscatter diffraction (EBSD), grain boundaries, misorientation, tessellation\n\nRelevant Snippet:\n\u2026 measured by 3D EBSD are investigated concerning segmentation, and \u2026 DREAM 3D software enable significant data evaluation that is completed by a five-parameter analysis of grain \u2026\n\n\n---\n\n## 4. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\nValid Text Names:\n'folwarczny2026accurategrainboundary pages 1-3', 'folwarczny2026accurategrainboundary pages 3-5', 'folwarczny2026accurategrainboundary pages 5-9', 'folwarczny2026accurategrainboundary pages 9-12', 'folwarczny2026accurategrainboundary pages 12-15', 'folwarczny2026accurategrainboundary pages 15-18', 'folwarczny2026accurategrainboundary pages 18-20'\n\nBibTex:\n@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n\n\nRelevant Snippet:\nWe present a method for obtaining qualitatively accurate grain boundary plane distributions (GBPD) for textured microstructures using a stereological calculation applied to two-\u2026\n\n\nUnobtainable Papers:\nRandi Swanson, Michael Chapman, Yue Zhou, Ashley Hilmas, Lisa Rueschhoff, Michael Uchic, William G. Fahrenholtz, and Scott J. McCormack. Quantitative grain structure and texture analysis of hot-pressed zrb2 via 3d ebsd. Materials & Design, 261:115378, Jan 2026. URL: https://doi.org/10.1016/j.matdes.2025.115378, doi:10.1016/j.matdes.2025.115378.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_TWgXFKMItXYFpWEbgg83FfU9", + "info": { + "start_ts": 625.591091176, + "end_ts": 664.734003185 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Automated Crystal Orientation Mapping in py4DSTEM using Sparse Correlation Matching\nValid Text Names:\n'ophus2022automatedcrystalorientation pages 1-2', 'ophus2022automatedcrystalorientation pages 2-4', 'ophus2022automatedcrystalorientation pages 4-5', 'ophus2022automatedcrystalorientation pages 5-6', 'ophus2022automatedcrystalorientation pages 6-7', 'ophus2022automatedcrystalorientation pages 7-8', 'ophus2022automatedcrystalorientation pages 8-10', 'ophus2022automatedcrystalorientation pages 10-11', 'ophus2022automatedcrystalorientation pages 11-12', 'ophus2022automatedcrystalorientation pages 12-13', 'ophus2022automatedcrystalorientation pages 13-14'\n\nBibTex:\n@article{ophus2022automatedcrystalorientation,\n author = \"Ophus, Colin and Zeltmann, Steven E and Bruefach, Alexandra and Rakowski, Alexander and Savitzky, Benjamin H and Minor, Andrew M and Scott, MC\",\n title = \"Automated Crystal Orientation Mapping in py4DSTEM using Sparse Correlation Matching\",\n year = \"2022\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"28\",\n pages = \"390-403\",\n month = \"Oct\",\n doi = \"10.48550/arxiv.2111.00171\",\n url = \"https://doi.org/10.48550/arxiv.2111.00171\",\n publisher = \"arXiv\"\n}\n\n\nAbstract:\nAbstract Crystalline materials used in technological applications are often complex assemblies composed of multiple phases and differently oriented grains. Robust identification of the phases and orientation relationships from these samples is crucial, but the information extracted from the diffraction condition probed by an electron beam is often incomplete. We have developed an automated crystal orientation mapping (ACOM) procedure which uses a converged electron probe to collect diffraction patterns from multiple locations across a complex sample. We provide an algorithm to determine the orientation of each diffraction pattern based on a fast sparse correlation method. We demonstrate the speed and accuracy of our method by indexing diffraction patterns generated using both kinematical and dynamical simulations. We have also measured orientation maps from an experimental dataset consisting of a complex polycrystalline twisted helical AuAgPd nanowire. From these maps we identify twin planes between adjacent grains, which may be responsible for the twisted helical structure. All of our methods are made freely available as open source code, including tutorials which can be easily adapted to perform ACOM measurements on diffraction pattern datasets.\n\nRelevant Snippet:\n\u2026 electron backscatter diffraction (EBSD) in scanning electron \u2026 disks in the pattern, then segmenting the data into radial bands\u2026 the orientation mapping application considered in this \u2026\n\n\n---\n\n## 2. High resolution crystal orientation mapping of ultrathin films in SEM and TEM\nValid Text Names:\n'heinig2022highresolutioncrystal pages 1-2', 'heinig2022highresolutioncrystal pages 2-3', 'heinig2022highresolutioncrystal pages 3-5', 'heinig2022highresolutioncrystal pages 5-6', 'heinig2022highresolutioncrystal pages 6-7', 'heinig2022highresolutioncrystal pages 7-8', 'heinig2022highresolutioncrystal pages 8-9', 'heinig2022highresolutioncrystal pages 9-9'\n\nBibTex:\n@misc{heinig2022highresolutioncrystal,\n author = \"Heinig, Mario F. and Chatterjee, Dipanwita and van Helvoort, Antonius T.J. and Wagner, Jakob Birkedal and Kadkhodazadeh, Shima and \u00c5nes, H\u00e5kon Wiik and Niessen, Frank and da Silva Fanta, Alice Bastos\",\n title = \"High resolution crystal orientation mapping of ultrathin films in SEM and TEM\",\n year = \"2022\",\n journal = \"Materials Characterization\",\n volume = \"189\",\n pages = \"111931\",\n month = \"Jul\",\n doi = \"10.1016/j.matchar.2022.111931\",\n url = \"https://doi.org/10.1016/j.matchar.2022.111931\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n\n\nRelevant Snippet:\nUltrathin metallic films are important functional materials for optical and microelectronic devices. Dedicated characterization with high spatial resolution and sufficient field of view is key \u2026\n\n\nUnobtainable Papers:\nDavid P. Field, Satyam Suwas, and Stuart I. Wright. Basics of SEM-Based Techniques for Electron Backscatter Diffraction, pages 27-41. Springer Nature Singapore, Jan 2025. URL: https://doi.org/10.1007/978-981-96-5346-1\\_2, doi:10.1007/978-981-96-5346-1\\_2.\nMarco Andres Acevedo Zamora, Christoph Eckart Schrank, and Balz Samuel Kamber. Using the traditional microscope for mineral grain orientation determination: a prototype image analysis pipeline for optic\u2010axis mapping (poam). Journal of Microscopy, 295:147-176, Mar 2024. URL: https://doi.org/10.1111/jmi.13284, doi:10.1111/jmi.13284.\nThomas Girerd, Giovanna Mart\u00ednez-Arellano, Adam Thomas Clare, and Alistair Speidel. A facile methodology to identify microstructural grains on etched surfaces using panoptic segmentation. Sep 2025. URL: https://doi.org/10.1016/j.matchar.2025.115224, doi:10.1016/j.matchar.2025.115224.\nYukun Liu, Patricia E. Meza, Zhi Li, Xiaobing Hu, Roberto dos Reis, Christopher Wolverton, and Vinayak Dravid. Seeing the invisible-understanding thermoelectric materials through scanning transmission electron microscopy. Chemistry of Materials, 38(8):3855-3878, Apr 2026. URL: https://doi.org/10.1021/acs.chemmater.5c02947, doi:10.1021/acs.chemmater.5c02947.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_oB8LA7Kpebnw5E9YNkkGkAve", + "info": { + "start_ts": 625.626175119, + "end_ts": 663.770777395 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nQiang Zhu, Chuanjie Wang, He-yong Qin, Gang Chen, and Peng Zhang. Effect of the grain size on the microtensile deformation and fracture behaviors of a nickel-based superalloy via ebsd and in-situ synchrotron radiation x-ray tomography. Materials Characterization, 156:109875, Oct 2019. URL: https://doi.org/10.1016/j.matchar.2019.109875, doi:10.1016/j.matchar.2019.109875.\nHao Chen, Yang Tao, Cheng Xin Li, and Han Xiao. The effect of grain size on the microstructural deformation behavior of cusn10p1 alloy under quasi-in-situ compression. Materials Science and Engineering: A, 973:150763, Oct 2026. URL: https://doi.org/10.1016/j.msea.2026.150763, doi:10.1016/j.msea.2026.150763.\nYaozhong Zhang, James Ball, Axel Henningsson, Jon Wright, Lucero Lopez, Dillon Jobes, Henry Proudhon, and Jerard V. Gordon. Unveiling 3d sub-grain residual stresses in as-built additively manufactured steel using scanning 3dxrd. Materials Research Letters, 13:700-708, May 2025. URL: https://doi.org/10.1080/21663831.2025.2502502, doi:10.1080/21663831.2025.2502502.\nMartin Diehl and Markus K\u00fchbach. Coupled experimental-computational analysis of primary static recrystallization in low carbon steel. Modelling and Simulation in Materials Science and Engineering, 28:014001, Nov 2019. URL: https://doi.org/10.1088/1361-651x/ab51bd, doi:10.1088/1361-651x/ab51bd.\nSven E Gustafson. Sub-grain deformation mechanisms of structural materials exposed through x-ray microscopy. Text, Jan 2023. URL: https://doi.org/10.25394/pgs.22151402, doi:10.25394/pgs.22151402.\njeremie bouquerel and Jean-Bernard VOGT. Assessment of ebsd regarding cyclic plasticity: a point of view. Unknown journal, Jan 2024. URL: https://doi.org/10.2139/ssrn.5077433, doi:10.2139/ssrn.5077433.\nLeo A.I. Kestens, Jesus Galan Lopez, Hadi Pirgazi, Tuan Nguyen-Minh, and Roumen H. Petrov. Texture observation and control in metal manufacturing : theory and practice. IOP Conference Series: Materials Science and Engineering, 1121:012002, Mar 2021. URL: https://doi.org/10.1088/1757-899x/1121/1/012002, doi:10.1088/1757-899x/1121/1/012002.\nZhijun Qiu, Hanliang Zhu, Zhiyang Wang, Ondrej Mur\u00e1nsky, Tao Wei, Elizabeth Budzakoska-Testone, Joel Davis, Andrew Studer, Q. F. Gu, Ulf Garbe, David Wexler, and Huijun Li. Crystallographic texture and multiscale boundaries mediated creep anisotropy in additively manufactured ni-based hastelloy c276 superalloy. Additive Manufacturing, Mar 2024. URL: https://doi.org/10.1016/j.addma.2024.104069, doi:10.1016/j.addma.2024.104069.\nSrinivas Doddapaneni, Sathish Kumar, Sathyashankara Sharma, Gowri Shankar, Manjunath Shettar, Nitesh Kumar, Ganesha Aroor, and Syed Mansoor Ahmad. Advancements in ebsd techniques: a comprehensive review on characterization of composites and metals, sample preparation, and operational parameters. Journal of Composites Science, 9:132, Mar 2025. URL: https://doi.org/10.3390/jcs9030132, doi:10.3390/jcs9030132.\nSoroosh Naghdy, L. L. Percq, R. Serret, Roumen Petrov, S. Hertel\u00e9, L. Kestens, and P. Verleysen. Microstructural evolution study of severely deformed commercial aluminium by transmission kikuchi diffraction. Materials Science and Technology, 33:678-687, Apr 2017. URL: https://doi.org/10.1080/02670836.2016.1194019, doi:10.1080/02670836.2016.1194019.\nF. Cruz-Gandarilla, \u2020. R.E.BOLMARO, \u2021. H.F.MENDOZA-LE\u00b4ON, A. M. S. J. C. \u02dcNAS-MORENO, F. Cruz-Gandarilla, Escuela Superior F\u00b4\u0131sica, Matem\u00b4aticas Instituto, Edificio Polit\u00b4ecnico Nacional, Del. G.A Madero, and C.P.07738. Ciudad de M\u00b4exico. Study of recovery and first recrystallisation kinetics in cgo fe3%si steels using misorientation\u2010derived parameters (ebsd). Journal of Microscopy, 275:133-148, Aug 2019. URL: https://doi.org/10.1111/jmi.12822, doi:10.1111/jmi.12822.\nJ. Collins, M. Taylor, A.L. Scarlett, E.J. Palmiere, and E.J. Pickering. Prior austenite grain measurement: a direct comparison of ebsd reconstruction, thermal etching and chemical etching. Materials Characterization, 208:113656, Feb 2024. URL: https://doi.org/10.1016/j.matchar.2024.113656, doi:10.1016/j.matchar.2024.113656.\n\u0160t\u011bp\u00e1n Gamanov, Poczkl\u00e1n Ladislav, Antonin Dlouhy, and Tomas Kruml. A step size dependence of kam, grod and gnd parameters calculated from ebsd. Jan 2025. URL: https://doi.org/10.2139/ssrn.4789333, doi:10.2139/ssrn.4789333.\nMoritz M\u00fcller, Amaya Compart, Mahnaz Mehdizadehlima, Serhii Yaroshevskyi, Christos G. Aneziris, Horst Biermann, and Anja Weidner. Ebsd study of creep\u2010induced lattice misorientation in mgo\u2010particle\u2010reinforced austenitic steel composites. Advanced Engineering Materials, Mar 2026. URL: https://doi.org/10.1002/adem.202502523, doi:10.1002/adem.202502523.\nFulin Wang, Jean-Charles Stinville, Marie Charpagne, McLean P. Echlin, Sean R. Agnew, Tresa M. Pollock, Marc De Graef, and Daniel S. Gianola. Dislocation cells in additively manufactured metallic alloys characterized by electron backscatter diffraction pattern sharpness. Mar 2023. URL: https://doi.org/10.1016/j.matchar.2023.112673, doi:10.1016/j.matchar.2023.112673.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_sXDuxIhQcjT0pGraCjqsLz2r", + "info": { + "start_ts": 625.665071324, + "end_ts": 664.716058467 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nJingwen Gao, Chenyang Zhu, Shubo Gao, Weiming Ji, Ming Xue, and Kun Zhou. Generative adversarial network\u2013enabled microstructural mapping from surface profiles for laser powder bed fusion. Virtual and Physical Prototyping, Jun 2025. URL: https://doi.org/10.1080/17452759.2025.2499925, doi:10.1080/17452759.2025.2499925.\nkikuchipy: an open-source toolbox for analysis of EBSD patterns\nJie Jiang, Xiaochen Li, and Zhenbo Zhang. Effect of scanning strategy on the formation of heterostructure and recrystallization behavior of additively manufactured in718 alloy. Jul 2026. URL: https://doi.org/10.1016/j.jmrt.2026.06.138, doi:10.1016/j.jmrt.2026.06.138.\nM. Fekih, T. Martinez Ostormujof, L. Germain, M. Piette, D. Sornin, and N. Gey. High temperature ebsd experiment versus crystallographic reconstruction to study phase transformation induced microstructures - application examples for steels. Materials Characterization, 228:115427, Oct 2025. URL: https://doi.org/10.1016/j.matchar.2025.115427, doi:10.1016/j.matchar.2025.115427.\nJinghua Cong, Shengjie Yang, Pengcheng Liu, Haibin Zhu, Xueliang Shang, Yongzhe Yang, and Xuemin Wang. Gradient temperature rolling induced homogenization of through-thickness mechanical properties and improved surface corrosion resistance in a heavy steel plate. Journal of Materials Research and Technology, Jun 2026. URL: https://doi.org/10.1016/j.jmrt.2026.06.009, doi:10.1016/j.jmrt.2026.06.009.\nIn Situ High Energy X-ray Diffraction Characterization of Phase Transformations and Mechanical Behaviors in Rapidly Solidified Titanium and Stainless Steel Alloys\nW.T. Zhang, R. Jiang, Y. Zhao, L.C. Zhang, L. Zhang, L.G. Zhao, and Y.D. Song. Effects of temperature and microstructure on low cycle fatigue behaviour of a pm ni-based superalloy: ebsd assessment and crystal plasticity simulation. International Journal of Fatigue, 159:106818, Jun 2022. URL: https://doi.org/10.1016/j.ijfatigue.2022.106818, doi:10.1016/j.ijfatigue.2022.106818.\nRong-lei Fan, Yong Wu, Minghe Chen, and Jie Zhao. Prediction of anisotropic deformation behavior of ta32 titanium alloy sheet during hot tension by crystal plasticity finite element model. May 2022. URL: https://doi.org/10.1016/j.msea.2022.143137, doi:10.1016/j.msea.2022.143137.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_I26KQDTI4oHtsQAWeLYGK9RU", + "info": { + "start_ts": 625.702459775, + "end_ts": 667.448846894 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures\nValid Text Names:\n'ostormujof2022deeplearningfor pages 1-3', 'ostormujof2022deeplearningfor pages 3-5', 'ostormujof2022deeplearningfor pages 5-9', 'ostormujof2022deeplearningfor pages 9-13', 'ostormujof2022deeplearningfor pages 13-16', 'ostormujof2022deeplearningfor pages 16-18', 'ostormujof2022deeplearningfor pages 18-21'\n\nBibTex:\n@article{ostormujof2022deeplearningfor,\n author = \"Ostormujof, T Martinez and Purohit, Rrp Purushottam Raj and Breumier, S and Gey, Nathalie and Salib, M and Germain, L\",\n title = \"Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures\",\n year = \"2022\",\n journal = \"Materials Characterization\",\n month = \"Nov\",\n doi = \"10.48550/arxiv.2112.03072\",\n url = \"https://doi.org/10.48550/arxiv.2112.03072\",\n publisher = \"arXiv\"\n}\n\n\nRelevant Snippet:\n\u2026 ) are able to extract meaningful features from EBSD-based data in order to automatically \u2026 parameters such as Kernel Average Misorientation (KAM) or pattern quality. Compared to other \u2026\n\n\n---\n\n## 2. Generalizable image segmentation for microstructure characterization through integrated SEM and EBSD analysis\nValid Text Names:\n'taufique2025generalizableimagesegmentation pages 1-2', 'taufique2025generalizableimagesegmentation pages 2-3', 'taufique2025generalizableimagesegmentation pages 3-4', 'taufique2025generalizableimagesegmentation pages 4-6', 'taufique2025generalizableimagesegmentation pages 6-7', 'taufique2025generalizableimagesegmentation pages 7-8', 'taufique2025generalizableimagesegmentation pages 8-10', 'taufique2025generalizableimagesegmentation pages 10-10'\n\nBibTex:\n@article{taufique2025generalizableimagesegmentation,\n author = \"Taufique, M. F. N. and Nguyen, Julia and Escobar, Julian D. and Pole, Mayur and Garcia, David and Wang, Tianhao and Das, Hrishikesh and Nwe, Kathy and Brown, David and Ross, Kenneth and Todd, Donald R. and Barker, Erin and Smith, Eric and Kappagantula, Keerti S. and Bilbrey, Jenna A.\",\n title = \"Generalizable image segmentation for microstructure characterization through integrated SEM and EBSD analysis\",\n year = \"2025\",\n journal = \"npj Computational Materials\",\n volume = \"11\",\n month = \"Oct\",\n doi = \"10.1038/s41524-025-01801-4\",\n url = \"https://doi.org/10.1038/s41524-025-01801-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n\n\nAbstract:\nAbstract\n We demonstrate generalizable semantic segmentation using minimal ground truth data. Correlated scanning electron microscopy (SEM) images and electron backscatter diffraction (EBSD) measurements of friction-stir processed 316L stainless steel plates were used to train deep learning models for grain boundary segmentation. Secondary electron (SE) imaging taken at 10\u2009keV correlated to EBSD-derived grain boundaries produced the best performing model. Notably, an ensemble of three models trained on a single SE image produced accurate segmentation over a series of backscatter electron (BSE) images of samples manufactured under different processing parameters, with a mean absolute error in grain size of 0.34\u2009\u00b5m. The generalizability of the models likely results from the similar escape depths of the SE training input and the EBSD training output and the reduced probability of stored strain artifacts appearing in the image. This highlights the importance of considering the physical principles behind imaging to develop robust models for microstructure characterization.\n\nRelevant Snippet:\n\u2026 seen by EBSD, which alters our perception of the ground truth \u2026 of the EBSD data, we reconstructed the grain boundaries \u2026 via kernel average misorientation (KAM) analysis. \u2026\n\n\nUnobtainable Papers:\nFederico Bruno, Georgios Konstantopoulos, Gianluca Fiore, Edoardo Rossi, Marco Sebastiani, Costas Charitidis, Luca Belforte, and Mauro Palumbo. A novel machine learning method to exploit ebsd and nanoindentation for trip steels microstructures analysis. Mar 2024. URL: https://doi.org/10.1016/j.matdes.2024.112774, doi:10.1016/j.matdes.2024.112774.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_5NDupt5Es4nrgT9SSGt0veGH", + "info": { + "start_ts": 625.742518522, + "end_ts": 662.189370401 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes\nValid Text Names:\n'polonsky2026pytribeamopensourcesoftware pages 1-2', 'polonsky2026pytribeamopensourcesoftware pages 2-5', 'polonsky2026pytribeamopensourcesoftware pages 5-7', 'polonsky2026pytribeamopensourcesoftware pages 7-8', 'polonsky2026pytribeamopensourcesoftware pages 8-10', 'polonsky2026pytribeamopensourcesoftware pages 10-12', 'polonsky2026pytribeamopensourcesoftware pages 12-14', 'polonsky2026pytribeamopensourcesoftware pages 14-15', 'polonsky2026pytribeamopensourcesoftware pages 15-16', 'polonsky2026pytribeamopensourcesoftware pages 16-17'\n\nBibTex:\n@article{polonsky2026pytribeamopensourcesoftware,\n author = \"Polonsky, Andrew T. and Lamb, James D. and Hovey, Chad B. and Schroader, Haydn and Echlin, McLean P. and Pollock, Tresa M.\",\n title = \"pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes\",\n year = \"2026\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"15\",\n pages = \"1-17\",\n month = \"Feb\",\n doi = \"10.1007/s40192-026-00449-2\",\n url = \"https://doi.org/10.1007/s40192-026-00449-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 the RAMS systems [15], and PyMicro based in situ testing [16]\u2026 Scientific (TFS) now have python APIs for microscope control, \u2026 , all EBSD parameters must be filled out in the EBSD vendor \u2026\n\n\n---\n\n## 2. Registration between DCT and EBSD datasets for multiphase microstructures\nValid Text Names:\n'ball2023registrationbetweendct pages 1-2', 'ball2023registrationbetweendct pages 2-4', 'ball2023registrationbetweendct pages 4-6', 'ball2023registrationbetweendct pages 6-7', 'ball2023registrationbetweendct pages 7-9', 'ball2023registrationbetweendct pages 9-11', 'ball2023registrationbetweendct pages 11-12', 'ball2023registrationbetweendct pages 12-14', 'ball2023registrationbetweendct pages 14-15', 'ball2023registrationbetweendct pages 15-15'\n\nBibTex:\n@article{ball2023registrationbetweendct,\n author = \"Ball, J. and Oddershede, J. and Davis, C. and Slater, C. and Said, M. and Vashishtha, H. and Michalik, \u0160. and of Metallurgy, David M. Collins School and Materials and Birmingham, U. and Kingdom., United and Ltd., Diamond Light Source and ApS, Xnovo Technology and Denmark and Wmg and Warwick, U.\",\n title = \"Registration between DCT and EBSD datasets for multiphase microstructures\",\n year = \"2023\",\n journal = \"Materials Characterization\",\n volume = \"204\",\n pages = \"113228\",\n month = \"Oct\",\n doi = \"10.1016/j.matchar.2023.113228\",\n url = \"https://doi.org/10.1016/j.matchar.2023.113228\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n\n\nRelevant Snippet:\n\u2026 the large grains for each phase present, austenite and martensitic ferrite. The results reveals a complex ferrite grain network of similar crystal orientations that are absent from the EBSD \u2026\n\n\nUnobtainable Papers:\nJames A. D. Ball, Anna Kareer, Oxana V. Magdysyuk, Stefan Michalik, Thomas Connolley, and David M. Collins. Per-grain and neighbourhood stress interactions during deformation of a ferritic steel obtained using three-dimensional x-ray diffraction. Text, Jan 2023. URL: https://doi.org/10.48550/arxiv.2302.10711, doi:10.48550/arxiv.2302.10711.\nYanling Schneider, Vighnesh Prabhu, Kai H\u00f6ss, Werner Wasserb\u00e4ch, Siegfried Schmauder, and Zhangjian Zhou. Many-scale investigations of the deformation behavior of polycrystalline composites: i\u2014machine learning applied for image segmentation. Materials, 15:2486, Mar 2022. URL: https://doi.org/10.3390/ma15072486, doi:10.3390/ma15072486.\nSangwon Lee, Michael Pilipchuk, C. Yildirim, Duncan A. Greeley, Qianying Shi, Tracy Berman, Adam Creuziger, Evan Rust, Carsten Detlefs, V. Sundararaghavan, John E. Allison, and Ashley Bucsek. Three-dimensional nucleation and growth of deformation twins in magnesium. Science, 389 6760:632-636, Dec 2025. URL: https://doi.org/10.48550/arxiv.2412.16640, doi:10.48550/arxiv.2412.16640.\nChristopher S. Daniel, Xiaohan Zeng, \u0160tefan Michalik, Simon A. Hunt, and Jo\u00e3o Quinta da Fonseca. Application of a new method for accurate determination of \u03b1 and \u03b2 texture in ti-6al-4v from synchrotron diffraction intensities. May 2023. URL: https://doi.org/10.1016/j.matchar.2023.112769, doi:10.1016/j.matchar.2023.112769.\nPierre Belamri, Henry Proudhon, Damien Texier, and David Ryckelynck. Quaternion-based vision-transformer for polycrystalline ebsd scans pre-trained on large-scale synthetic data. Materials & Design, 258:114599, Oct 2025. URL: https://doi.org/10.1016/j.matdes.2025.114599, doi:10.1016/j.matdes.2025.114599.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0000 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_TcPaoKE5klpaxx5d8PlBUDrk", + "info": { + "start_ts": 625.798862717, + "end_ts": 671.189819695 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_YCK1KFOsT4kHD9IBqs5QYRHi", + "type": "function", + "function": { + "arguments": "{\"question\": \"Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "No relevant evidence found for your question.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=12 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_YCK1KFOsT4kHD9IBqs5QYRHi", + "info": { + "start_ts": 681.740740548, + "end_ts": 688.630316028 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_YAEYGtulkKwxKg8CZdFRkuKJ", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns\\\"\"}", + "name": "paper_search" + } + }, + { + "id": "call_6EDjDZUx9EcloPKavvbsh1xb", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"orix\\\" Python crystallographic orientation library\"}", + "name": "paper_search" + } + }, + { + "id": "call_bd44zev0ajEnct80hGQyaouJ", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"DefDAP\\\" deformation data analysis Python\"}", + "name": "paper_search" + } + }, + { + "id": "call_Z5I4Dszqv8IduAWdkGxzojFr", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"DREAM.3D\\\" \\\"Integrating Materials and Manufacturing Innovation\\\" software\"}", + "name": "paper_search" + } + }, + { + "id": "call_5hiGgVfNiBeJ5xEkpggCPT0O", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"MTEX\\\" quantitative texture analysis software\"}", + "name": "paper_search" + } + }, + { + "id": "call_t3X9GkiKh5YtIwvvFGFtqFju", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"PyMicro\\\" Python microstructure library crystallography\"}", + "name": "paper_search" + } + }, + { + "id": "call_2WFpaGkkorc7QNUjgoGtFJOi", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"pyxem\\\" open-source Python library diffraction microscopy\"}", + "name": "paper_search" + } + }, + { + "id": "call_xqX7V5hmrKT4tjzjK7lLSxlE", + "type": "function", + "function": { + "arguments": "{\"query\": \"\\\"OpenXY\\\" EBSD Python GND\"}", + "name": "paper_search" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_YAEYGtulkKwxKg8CZdFRkuKJ", + "info": { + "start_ts": 703.386594257, + "end_ts": 745.729313749 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nD. Johnstone, Ben H. Martineau, P. Crout, P. Midgley, and A. Eggeman. Density-based clustering of crystal (mis)orientations and the orix python library. Journal of Applied Crystallography, 53:1293-1298, Sep 2020. URL: https://doi.org/10.1107/s1600576720011103, doi:10.1107/s1600576720011103.\nIan MacLaren, Enrique Frutos\u2010Myro, Steven Zeltmann, and Colin Ophus. A method for crystallographic mapping of an alpha\u2010beta titanium alloy with nanometre resolution using scanning precession electron diffraction and open\u2010source software libraries. Journal of Microscopy, 295:131-139, Feb 2024. URL: https://doi.org/10.1111/jmi.13275, doi:10.1111/jmi.13275.\nZhuo Li, Daren Wu, Simon Si Ming Ji, Ana G. Claus, Hui Zhong, Sanjit K. Ghose, and Kelsey B. Hatzell. Texture evolution of plated lithium in anode-free solid-state batteries. ACS Energy Letters, 11(2):1705-1713, Jan 2026. URL: https://doi.org/10.1021/acsenergylett.5c03205, doi:10.1021/acsenergylett.5c03205.\nDorian Depriester and R\u00e9gis Kubler. Elasticipy: a python package for linear elasticity and tensor analysis. Journal of Open Source Software, 10:7940, Nov 2025. URL: https://doi.org/10.21105/joss.07940, doi:10.21105/joss.07940.\nMattis Seehaus, Sang-Hyeok Lee, Tobias Stollenwerk, Jeffrey M. Wheeler, and Sandra Korte-Kerzel. Estimation of directional single crystal elastic properties from nano-indentation by correlation with ebsd and first-principle calculations. Materials & Design, 234:112296, Oct 2023. URL: https://doi.org/10.1016/j.matdes.2023.112296, doi:10.1016/j.matdes.2023.112296.\nZachary Varley. Algorithms for crystallography in the scanning electron microscope. Text, Jan 2024. URL: https://doi.org/10.1184/r1/26955991, doi:10.1184/r1/26955991.\nColin Ophus. Quantitative scanning transmission electron microscopy for materials science: imaging, diffraction, spectroscopy, and tomography. Annual Review of Materials Research, 53:105-141, Jul 2023. URL: https://doi.org/10.1146/annurev-matsci-080921-092646, doi:10.1146/annurev-matsci-080921-092646.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_6EDjDZUx9EcloPKavvbsh1xb", + "info": { + "start_ts": 703.425128648, + "end_ts": 746.329071205 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nBenjamin Poole, Alex Marsh, David Lunt, Chris Hardie, Mike Gorley, Cory Hamelin, and Allan Harte. High\u2010resolution strain mapping in a thermionic lab6 scanning electron microscope. Strain, Feb 2024. URL: https://doi.org/10.1111/str.12472, doi:10.1111/str.12472.\nYangchao Deng, Zhenjie Cao, Dongchen Hu, Albert D. Smith, Richard Hunt, Kun Yan, and Joseph D. Robson. Insights into strain partitioning and void nucleation in recycled aluminium alloy: the role of fe-rich intermetallic. Materials Science and Engineering: A, 973:150757, Oct 2026. URL: https://doi.org/10.1016/j.msea.2026.150757, doi:10.1016/j.msea.2026.150757.\nConghui Liu, Xu Xu, Tianzhu Sun, Rhys Thomas, Jo\u00e3o Quinta da Fonseca, and Michael Preuss. Microstructural effects on fatigue crack initiation mechanisms in a near-alpha titanium alloy. Jul 2023. URL: https://doi.org/10.1016/j.actamat.2023.118957, doi:10.1016/j.actamat.2023.118957.\nRhys Thomas, David Bowden, David Lunt, David Stewart, and Michael Preuss. High compressive loading performance of a complex multi-phase hard-facing alloy explained through a highly elastic silicide phase. Aug 2023. URL: https://doi.org/10.1016/j.mtla.2023.101815, doi:10.1016/j.mtla.2023.101815.\nMichael D Atkinson, Rhys Thomas, Allan Harte, Peter Crowther, and Jo\u00e3o Quinta da Fonseca. Defdap: deformation data analysis in python. Feb 2020. URL: https://doi.org/10.5281/zenodo.3688097, doi:10.5281/zenodo.3688097.\nConghui Liu, Tianzhu Sun, Dongchen Hu, Rhys Thomas, Philipp Frankel, Jo\u00e3o Quinta da Fonseca, and Michael Preuss. Mechanistic insight into cooperative slip system activation under cyclic loading in a near-alpha titanium alloy. Acta Materialia, 308:122031, Apr 2026. URL: https://doi.org/10.1016/j.actamat.2026.122031, doi:10.1016/j.actamat.2026.122031.\nRan Ni, Carl J. Boehlert, Bo Chen, Yuanshuai Zhang, Ying Zeng, J. Zheng, Hao Zhou, Qudong Wang, and Dongdi Yin. Temperature-dependent interplay of intra- and inter-granular deformation mechanisms in mg-10y: statistical analysis from an hrdic perspective. Acta Materialia, Jun 2025. URL: https://doi.org/10.1016/j.actamat.2025.121256, doi:10.1016/j.actamat.2025.121256.\nSunil Anandatheertha, Vikram Phalke, Eralp Demir, and Christopher Hardie. Ukaea poly-xtal operations (upxo v1.0): an open-source python package for generating, assessment and meshing of poly-crystalline grain structures. SoftwareX, 34:102736, Jun 2026. URL: https://doi.org/10.1016/j.softx.2026.102736, doi:10.1016/j.softx.2026.102736.\nDongchen Hu, Albert D. Smith, David Lunt, Rhys Thomas, Michael D. Atkinson, Xiaodong Liu, \u00d6mer Ko\u00e7, Jack M. Donoghue, Zhenbo Zhang, Jo\u00e3o Quinta da Fonseca, and Michael Preuss. Tracking the onset of plasticity in a ni-base superalloy using in-situ high-resolution digital image correlation. Feb 2025. URL: https://doi.org/10.1016/j.matchar.2024.114654, doi:10.1016/j.matchar.2024.114654.\nBerzah Yavuzyegit, Egemen Avcu, Albert D. Smith, Jack M. Donoghue, David Lunt, Joseph D. Robson, Timothy L. Burnett, Jo\u00e3o Quinta da Fonseca, and Philip J. Withers. Mapping plastic deformation mechanisms in az31 magnesium alloy at the nanoscale. SSRN Electronic Journal, 250:118876, May 2023. URL: https://doi.org/10.2139/ssrn.4198880, doi:10.2139/ssrn.4198880.\nJie Luo, Albert D. Smith, Jack M. Donoghue, Jiaqi Xu, Juan Nunez-Iglesias, Christopher Hutchinson, Philip J. Withers, and Michael Preuss. Automated in-situ testing of deformation and phase transformations in engineering alloys. Automated In-situ Testing of Deformation and Phase transformations in Engineering Alloys, Jan 2025. URL: https://doi.org/10.14293/apmc13-2025-0293, doi:10.14293/apmc13-2025-0293.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_bd44zev0ajEnct80hGQyaouJ", + "info": { + "start_ts": 703.459767095, + "end_ts": 740.862397661 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\nValid Text Names:\n'polonsky2019threedimensionalanalysisand pages 1-2', 'polonsky2019threedimensionalanalysisand pages 2-5', 'polonsky2019threedimensionalanalysisand pages 5-6', 'polonsky2019threedimensionalanalysisand pages 6-8', 'polonsky2019threedimensionalanalysisand pages 8-11', 'polonsky2019threedimensionalanalysisand pages 11-13', 'polonsky2019threedimensionalanalysisand pages 13-14', 'polonsky2019threedimensionalanalysisand pages 14-15', 'polonsky2019threedimensionalanalysisand pages 15-15'\n\nBibTex:\n@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a cloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline variables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore, a comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using BisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects for impact in material science are also discussed.\n\nRelevant Snippet:\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a cloud-based platform, BisQue. Parallelization of DREAM.3D module \u2026\n\n\n---\n\n## 2. ITKMontage: A Software Module for Image Stitching\nValid Text Names:\n'zukic2021itkmontageasoftware pages 1-2', 'zukic2021itkmontageasoftware pages 2-4', 'zukic2021itkmontageasoftware pages 4-5', 'zukic2021itkmontageasoftware pages 5-8', 'zukic2021itkmontageasoftware pages 8-10', 'zukic2021itkmontageasoftware pages 10-10'\n\nBibTex:\n@article{zukic2021itkmontageasoftware,\n author = \"Zuki\u0107, D\u017eenan and Jackson, Michael and Dimiduk, Dennis and Donegan, Sean and Groeber, Michael and McCormick, Matthew\",\n title = \"ITKMontage: A Software Module for Image Stitching\",\n year = \"2021\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"10\",\n pages = \"115-124\",\n month = \"Mar\",\n doi = \"10.1007/s40192-021-00202-x\",\n url = \"https://doi.org/10.1007/s40192-021-00202-x\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 Its liberal Apache 2.0 license allows incorporation of the code into a wide range of software projects, including proprietary software. We present here a montaging software package \u2026\n\n\n---\n\n## 3. Framework for the Generation of 3D Fiber Composite Structures from 2D Observations\nValid Text Names:\n'clarke2024frameworkforthe pages 1-2', 'clarke2024frameworkforthe pages 2-4', 'clarke2024frameworkforthe pages 4-6', 'clarke2024frameworkforthe pages 6-7', 'clarke2024frameworkforthe pages 7-9', 'clarke2024frameworkforthe pages 9-13', 'clarke2024frameworkforthe pages 13-15', 'clarke2024frameworkforthe pages 15-16', 'clarke2024frameworkforthe pages 16-16'\n\nBibTex:\n@article{clarke2024frameworkforthe,\n author = \"Clarke, Kenneth M. and Wertz, John and Groeber, Michael\",\n title = \"Framework for the Generation of 3D Fiber Composite Structures from 2D Observations\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"13\",\n pages = \"569-584\",\n month = \"May\",\n doi = \"10.1007/s40192-024-00352-8\",\n url = \"https://doi.org/10.1007/s40192-024-00352-8\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nSynthetic generation of realistic materials for testing of process\u2013structure\u2013property (PSP) relationships in computational materials science has gained significant traction over the past two decades. Generation tools continue to lag in some aspects of realism, leading to uncertainty and errors in simulating material response. The experimental collection of information to guide generation of 3D synthetic structures remains costly, time consuming, and challenging. These challenges are compounded by limitations of stereology, which permits estimation of 3D microstructural statistics from 2D observations under restrictive assumptions on constituent morphologies and size distributions and can be difficult to apply to microstructure metrics like constituent orientation distribution and clustering. This work seeks to overcome these challenges by introducing a framework for learning probable 3D microstructure statistics by minimizing the loss determined by matching statistics obtained from 2D observations via synthetic microstructure generation software (e.g. Dream.3D) and stereological principles. This framework is applied to short carbon fiber composite structures printed from an additive manufacturing process, Direct Ink Writing, with the hope that the framework could generalize to other particulate structures as well as allow for simulation and design of materials structure.\n\nRelevant Snippet:\n\u2026 Structures from 2D Observations | Integrating Materials and Manufacturing Innovation \u2026 from 2D observations via synthetic microstructure generation software (eg Dream.3D) and \u2026\n\n\n---\n\n## 4. AFRL Additive Manufacturing Modeling Series: Challenge 4, 3D Reconstruction of an IN625 High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning\nValid Text Names:\n'chapman2021afrladditivemanufacturing pages 1-3', 'chapman2021afrladditivemanufacturing pages 3-5', 'chapman2021afrladditivemanufacturing pages 5-7', 'chapman2021afrladditivemanufacturing pages 7-8', 'chapman2021afrladditivemanufacturing pages 8-10', 'chapman2021afrladditivemanufacturing pages 10-11', 'chapman2021afrladditivemanufacturing pages 11-13', 'chapman2021afrladditivemanufacturing pages 13-13'\n\nBibTex:\n@article{chapman2021afrladditivemanufacturing,\n author = \"Chapman, Michael G. and Shah, Megna N. and Donegan, Sean P. and Scott, J. Michael and Shade, Paul A. and Menasche, David and Uchic, Michael D.\",\n title = \"AFRL Additive Manufacturing Modeling Series: Challenge 4, 3D Reconstruction of an IN625 High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning\",\n year = \"2021\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"10\",\n pages = \"129-141\",\n month = \"Jun\",\n doi = \"10.1007/s40192-021-00212-9\",\n url = \"https://doi.org/10.1007/s40192-021-00212-9\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n\n\nAbstract:\nHigh-energy diffraction microscopy (HEDM) in-situ mechanical testing experiments offer unique insight into the evolving deformation state within polycrystalline materials. These experiments rely on a sophisticated analysis of the diffraction data to instantiate a 3D reconstruction of grains and other microstructural features associated with the test volume. For microstructures of engineering alloys that are highly twinned and contain numerous features around the estimated spatial resolution of HEDM reconstructions, the accuracy of the reconstructed microstructure is not known. In this study, we address this uncertainty by characterizing the same HEDM sample volume using destructive serial sectioning (SS) that has higher spatial resolution. The SS experiment was performed on an Inconel 625 alloy sample that had undergone HEDM in-situ mechanical testing to a small amount of plastic strain (~\u20090.7%), which was part of the Air Force Research Laboratory Additive Manufacturing (AM) Modeling Series. A custom-built automated multi-modal SS system was used to characterize the entire test volume, with a spatial resolution of approximately 1 \u00b5m. Epi-illumination optical microscopy images, backscattered electron images, and electron backscattered diffraction maps were collected on every section. All three data modes were utilized and custom data fusion protocols were developed for 3D reconstruction of the test volume. The grain data were homogenized and downsampled to 2 \u00b5m as input for Challenge 4 of the AM Modeling Series, which is available at the Materials Data Facility repository.\n\nRelevant Snippet:\n\u2026 3D file, and can be viewed with any software program that is capable of reading and viewing an HDF5 file. In addition, if the text in the \u2018Pipeline\u2019 group is saved as a new *.json file, this \u2026\n\n\n---\n\n## 5. AixViPMaP\u00ae\u2014an Operational Platform for Microstructure Modeling Workflows\nValid Text Names:\n'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 1-2', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 2-3', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 3-5', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 5-8', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 8-11', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 11-12', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 12-14', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 14-17', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 17-19', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 19-21', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 21-22', 'koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform pages 22-22'\n\nBibTex:\n@article{koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform,\n author = \"Koschmieder, L. and Hojda, S. and Apel, M. and Altenfeld, R. and Bami, Y. and Haase, C. and Lin, M. and Vuppala, A. and Hirt, G. and Schmitz, G.J.\",\n title = \"AixViPMaP\u00ae\u2014an Operational Platform for Microstructure Modeling Workflows\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"122-143\",\n month = \"Jun\",\n doi = \"10.1007/s40192-019-00138-3\",\n url = \"https://doi.org/10.1007/s40192-019-00138-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n\n\nRelevant Snippet:\n\u2026 supporting some but not all simulation software. For a co-simulation framework to support software, usually, the data model of the specific software must be specified and additional APIs \u2026\n\n\nUnobtainable Papers:\nR Brian Leavy, John Clayton, Jonathan Ligda, and Jaroslaw Knap. Methods for microstructure modeling and phase field simulations of fracture. ArXiv, Jun 2023. URL: https://doi.org/10.21236/ad1203858, doi:10.21236/ad1203858.\nG. J. Schmitz. Microstructure modeling in integrated computational materials engineering (icme) settings: can hdf5 provide the basis for an emerging standard for describing microstructures? JOM, 68:77-83, Dec 2016. URL: https://doi.org/10.1007/s11837-015-1748-2, doi:10.1007/s11837-015-1748-2.\nStanislav Buklovskyi, Kateryna Miroshnichenko, Igor Tsukrov, Rebecca J. Thomson, Peder C. Solberg, and Douglas W. Van Citters. Effective thermal conductivity of carbon/ultra-high-molecular-weight-polyethylene nanocomposites. Procedia Structural Integrity, 80:146-156, Jan 2026. URL: https://doi.org/10.1016/j.prostr.2026.02.014, doi:10.1016/j.prostr.2026.02.014.\nCarter K. Cocke, Anthony D. Rollett, Ricardo A. Lebensohn, and Ashley D. Spear. The afrl additive manufacturing modeling challenge: predicting micromechanical fields in am in625 using an fft-based method with direct input from a 3d microstructural image. Integrating Materials and Manufacturing Innovation, 10:157-176, May 2021. URL: https://doi.org/10.1007/s40192-021-00211-w, doi:10.1007/s40192-021-00211-w.\nAlexander Ra\u00dfloff, Paul Schulz, Robert K\u00fchne, Marreddy Ambati, Ilja Koch, Andr\u00e9 T. Zeuner, Maik Gude, Martina Zimmermann, and Markus K\u00e4stner. Experimental-numerical analysis of microstructure-property linkages for additively manufactured materials. Procedia Structural Integrity, 38:4-11, Jan 2022. URL: https://doi.org/10.1016/j.prostr.2022.03.002, doi:10.1016/j.prostr.2022.03.002.\nM. Jackson, M. Groeber, S. Donegan, and D. Dimiduk. Advancements to the digital representation environment for analysis of materials in 3-dimensions\u2014dream.3d. Microscopy and Microanalysis, 25:176-177, Aug 2019. URL: https://doi.org/10.1017/s1431927619001612, doi:10.1017/s1431927619001612.\nJonathan Robichaud, Tim Vincent, Ben Schultheis, and Anil Chaudhary. Integrated computational materials engineering to predict melt-pool dimensions and 3d grain structures for selective laser melting of inconel 625. Integrating Materials and Manufacturing Innovation, 8:305-317, Jul 2019. URL: https://doi.org/10.1007/s40192-019-00145-4, doi:10.1007/s40192-019-00145-4.\nAyman A. Salem, Joshua B. Shaffer, Richard A. Kublik, Luke A. Wuertemberger, and Daniel P. Satko. Microstructure-informed cloud computing for interoperability of materials databases and computational models: microtextured regions in ti alloys. Mar 2017. URL: https://doi.org/10.1007/s40192-017-0090-7, doi:10.1007/s40192-017-0090-7.\nJohn Wertz, Erik P. Blasch, Laura Homa, and Matt Cherry. Dynamic Data Driven Applications Systems Analysis of Microtexture Regions in Titanium Alloys, pages 199-222. Springer Nature Switzerland, Jan 2026. URL: https://doi.org/10.1007/978-3-031-88574-7\\_7, doi:10.1007/978-3-031-88574-7\\_7.\nBin Li, Tao Gao, Zhi Qin, Hongqian Xue, and Zhidan Sun. Investigation of the fatigue life scatter for aa7075-t6 using crystal plasticity finite element method in the high to very high cycle fatigue regime. Integrating Materials and Manufacturing Innovation, 11:198-213, Mar 2022. URL: https://doi.org/10.1007/s40192-022-00256-5, doi:10.1007/s40192-022-00256-5.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_Z5I4Dszqv8IduAWdkGxzojFr", + "info": { + "start_ts": 703.492846157, + "end_ts": 739.275643191 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\nValid Text Names:\n'savage2021anautomatedprocedure pages 1-2', 'savage2021anautomatedprocedure pages 2-3', 'savage2021anautomatedprocedure pages 3-4', 'savage2021anautomatedprocedure pages 4-5', 'savage2021anautomatedprocedure pages 5-6', 'savage2021anautomatedprocedure pages 6-7', 'savage2021anautomatedprocedure pages 7-9', 'savage2021anautomatedprocedure pages 9-11', 'savage2021anautomatedprocedure pages 11-13', 'savage2021anautomatedprocedure pages 13-14', 'savage2021anautomatedprocedure pages 14-15', 'savage2021anautomatedprocedure pages 15-16', 'savage2021anautomatedprocedure pages 16-16'\n\nBibTex:\n@article{savage2021anautomatedprocedure,\n author = \"Savage, Daniel J. and McCabe, Rodney J. and Knezevic, Marko\",\n title = \"An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\",\n volume = \"171\",\n issn = \"1044-5803\",\n url = \"https://doi.org/10.1016/j.matchar.2020.110808\",\n doi = \"10.1016/j.matchar.2020.110808\",\n journal = \"Materials Characterization\",\n publisher = \"Elsevier BV\",\n year = \"2021\",\n month = \"Jan\",\n pages = \"110808\"\n}\n\n\nRelevant Snippet:\n\u2026 or neutron diffraction bulk texture measurements based on the quantitative change in fraction of c-axes within some angle to the compression direction [17, 53]. The texture method is \u2026\n\n\n---\n\n## 2. A Survey and Comparison of Methods for Quantitative TextureAnalysis\nValid Text Names:\n'einhorn2017asurveyand pages 1-5', 'einhorn2017asurveyand pages 5-7', 'einhorn2017asurveyand pages 7-11', 'einhorn2017asurveyand pages 11-16', 'einhorn2017asurveyand pages 16-20', 'einhorn2017asurveyand pages 20-23', 'einhorn2017asurveyand pages 23-27', 'einhorn2017asurveyand pages 27-30', 'einhorn2017asurveyand pages 30-35', 'einhorn2017asurveyand pages 35-41', 'einhorn2017asurveyand pages 41-46', 'einhorn2017asurveyand pages 46-52', 'einhorn2017asurveyand pages 52-56', 'einhorn2017asurveyand pages 56-60', 'einhorn2017asurveyand pages 60-63', 'einhorn2017asurveyand pages 67-71', 'einhorn2017asurveyand pages 71-75', 'einhorn2017asurveyand pages 79-84', 'einhorn2017asurveyand pages 88-92', 'einhorn2017asurveyand pages 92-95', 'einhorn2017asurveyand pages 95-98', 'einhorn2017asurveyand pages 98-102', 'einhorn2017asurveyand pages 102-104', 'einhorn2017asurveyand pages 104-107', 'einhorn2017asurveyand pages 109-109'\n\nBibTex:\n@article{einhorn2017asurveyand,\n author = \"Einhorn, John\",\n title = \"A Survey and Comparison of Methods for Quantitative TextureAnalysis\",\n year = \"2017\",\n journal = \"ArXiv\",\n doi = \"10.18130/v30r4d\",\n url = \"https://doi.org/10.18130/v30r4d\",\n month = \"Jul\",\n publisher = \"University of Virginia\"\n}\n\n\nAbstract:\nNumerous methods for measuring and analyzing texture\nare explored and discussed. Traditional pole figure inversion\nmethods are compared against more recent developments in\nquantitative texture analysis involving the Rietveld whole pattern\nrefinement method. A modified equal area pole figure collection\nstrategy is compared to the conventional equiangular method using\nlaboratory X-ray diffraction (XRD). The two measurement strategies\nwere found to yield qualitatively similar orientation distribution\nfunctions (ODF), and the former set takes about half the time to\ncollect. However, the estimated texture was found to be strongly\ndependent on the ODF computation parameters used in MTEX. The\ntexture evolution during compression of a high specific strength\nMg-base alloy containing long period stacking order (LPSO) phase is\nstudied using synchrotron XRD. The initial texture was found to be\nmuch sharper than previously published results, demonstrating the\nimportance of using a \u201cdirect method\u201d to compute the OD for this\nmaterial. The evolution of the texture of the Mg and LPSO phases\nsuggests that the deformation is accommodated almost entirely by\ntwinning in the former case and by dislocation slip, elastic\ndeformation, and possibly some twinning in the latter. The texture\nof \u03b1-uranium foils is measured via X-ray (both Cu- and Mo-K\u03b1\nradiation) and neutron diffraction, and the results are analyzed\nwith MTEX and MAUD (Rietveld), respectively. The three radiation\nsources are found to yield slightly different results, based on\nprobing different depths of the foils. Finally, the capability of\nmeasuring texture on the Neutron Residual Stress Mapping Facility\n(NRSF2) and Nanoscale Ordered Materials Diffractometer (NOMAD)\ninstruments at the Oak Ridge National Laboratory (ORNL) is\ndemonstrated by comparing measurements to those obtained via\nconventional XRD methods. Pole figure inversion in MTEX (for XRD\nand NRSF2) and Rietveld analysis in MAUD (for NOMAD) are employed\nfor analysis, and the results are quantitatively compared.\nReasonable agreement is found across all methods, and the\ndifferences between methods are attributed mostly to differences in\nthe analysis procedure. The implications of slight variations in\ntexture, as estimated by the various techniques, are explored\nthrough calculations of tensor properties and comparisons with\npredicted deformation texture evolutions. Finally, suggestions for\nimproving the data collection speed by collecting less data are\nexplored for both instruments.\n\nRelevant Snippet:\n\u2026 However, the estimated texture was found to be strongly dependent on the ODF computation parameters used in MTEX. The texture evolution during compression of a high specific \u2026\n\n\n---\n\n## 3. Quantitative texture analysis and combined analysis\nValid Text Names:\n'chateigner2019quantitativetextureanalysis pages 1-2', 'chateigner2019quantitativetextureanalysis pages 2-3', 'chateigner2019quantitativetextureanalysis pages 3-4', 'chateigner2019quantitativetextureanalysis pages 4-5', 'chateigner2019quantitativetextureanalysis pages 5-6', 'chateigner2019quantitativetextureanalysis pages 6-7', 'chateigner2019quantitativetextureanalysis pages 7-8', 'chateigner2019quantitativetextureanalysis pages 8-10', 'chateigner2019quantitativetextureanalysis pages 10-11', 'chateigner2019quantitativetextureanalysis pages 11-12', 'chateigner2019quantitativetextureanalysis pages 12-13', 'chateigner2019quantitativetextureanalysis pages 13-15', 'chateigner2019quantitativetextureanalysis pages 15-16', 'chateigner2019quantitativetextureanalysis pages 16-17', 'chateigner2019quantitativetextureanalysis pages 17-18', 'chateigner2019quantitativetextureanalysis pages 18-19', 'chateigner2019quantitativetextureanalysis pages 18-18', 'chateigner2019quantitativetextureanalysis pages 19-21', 'chateigner2019quantitativetextureanalysis pages 21-22', 'chateigner2019quantitativetextureanalysis pages 22-23', 'chateigner2019quantitativetextureanalysis pages 23-24', 'chateigner2019quantitativetextureanalysis pages 24-25', 'chateigner2019quantitativetextureanalysis pages 25-26', 'chateigner2019quantitativetextureanalysis pages 26-26'\n\nBibTex:\n@inbook{chateigner2019quantitativetextureanalysis,\n author = \"Chateigner, D. and Lutterotti, L. and Morales, M.\",\n title = \"Quantitative texture analysis and combined analysis\",\n ISBN = \"9781118416280\",\n url = \"https://doi.org/10.1107/97809553602060000968\",\n doi = \"10.1107/97809553602060000968\",\n booktitle = \"International Tables for Crystallography\",\n publisher = \"International Union of Crystallography\",\n year = \"2019\",\n month = \"Dec\",\n pages = \"555-580\",\n journal = \"International Tables for Crystallography\"\n}\n\n\nRelevant Snippet:\n\u2026 Here, we illustrate magnetic quantitative texture analysis on an iron sample exhibiting magnetic saturation around 0.5T (Chateigner et al., 2010) that was initially not magnetized. \u2026\n\n\nUnobtainable Papers:\nS. Gorelova, H. Schaeben, and R. Kawalla. Quantifying texture evolution during hot rolling of magnesium twin roll cast strip. Materials Science and Engineering A-structural Materials Properties Microstructure and Processing, 602:134-142, Apr 2014. URL: https://doi.org/10.1016/j.msea.2014.02.057, doi:10.1016/j.msea.2014.02.057.\nFlorencia Malamud, Steve Gaudez, Jan Capek, Ezequiel Fogliatto, and Markus Strobl. Texture analysis implementation at the neutron strain diffractometer poldi. Materials Characterization, Apr 2025. URL: https://doi.org/10.1016/j.matchar.2025.115003, doi:10.1016/j.matchar.2025.115003.\nS. Gorelova and H. Schaeben. Quantifying texture evolution during hot rolling of az31 twin roll cast strip. IOP Conference Series: Materials Science and Engineering, 82:012003, Apr 2015. URL: https://doi.org/10.1088/1757-899x/82/1/012003, doi:10.1088/1757-899x/82/1/012003.\nGennady Rafailov, El'ad N. Caspi, Ralf Hielscher, Eitan Tiferet, Roni Schneck, and Sven C. Vogel. Visualization of texture components using mtex. Journal of Applied Crystallography, 53:540-548, Feb 2020. URL: https://doi.org/10.1107/s1600576719014742, doi:10.1107/s1600576719014742.\nRoberto Visalli, Gaetano Ortolano, Gaston Godard, and Rosolino Cirrincione. Micro-fabric analyzer (mfa): a new semiautomated arcgis-based edge detector for quantitative microstructural analysis of rock thin-sections. ISPRS Int. J. Geo Inf., 10:51, Jan 2021. URL: https://doi.org/10.3390/ijgi10020051, doi:10.3390/ijgi10020051.\nDavid Mainprice, Ralf Hielscher, and Helmut Schaeben. Calculating anisotropic physical properties from texture data using the mtex open-source package. Geological Society, London, Special Publications, 360:175-192, Jan 2011. URL: https://doi.org/10.1144/sp360.10, doi:10.1144/sp360.10.\nFlorian Bachmann, Ralf Hielscher, and Helmut Schaeben. Grain detection from 2d and 3d ebsd data\u2013specification of the mtex algorithm. Ultramicroscopy, 111 12:1720-33, Dec 2011. URL: https://doi.org/10.1016/j.ultramic.2011.08.002, doi:10.1016/j.ultramic.2011.08.002.\nM. Y. Xie, N. Baimpas, C. Reinhard, and A. M. Korsunsky. Texture analysis in cubic phase polycrystals by single exposure synchrotron x-ray diffraction. Journal of Applied Physics, 114:163502, Oct 2013. URL: https://doi.org/10.1063/1.4825120, doi:10.1063/1.4825120.\nV. A. Yardley, S. Fahimi, and E. J. Payton. Classification of creep crack and cavitation sites in tempered martensite ferritic steel microstructures using mtex toolbox for ebsd. Materials Science and Technology, 31:547-553, Mar 2015. URL: https://doi.org/10.1179/1743284714y.0000000603, doi:10.1179/1743284714y.0000000603.\nDavid Mainprice, Florian Bachmann, Ralf Hielscher, and Helmut Schaeben. Descriptive tools for the analysis of texture projects with large datasets using mtex: strength, symmetry and components. Geological Society, London, Special Publications, 409:251-271, Aug 2015. URL: https://doi.org/10.1144/sp409.8, doi:10.1144/sp409.8.\nDavid Mainprice, Florian Bachmann, Ralf Hielscher, Helmut Schaeben, and Geoffrey E. Lloyd. Calculating anisotropic piezoelectric properties from texture data using the mtex open source package. Geological Society, London, Special Publications, 409:223-249, Jul 2015. URL: https://doi.org/10.1144/sp409.2, doi:10.1144/sp409.2.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_5hiGgVfNiBeJ5xEkpggCPT0O", + "info": { + "start_ts": 703.531510634, + "end_ts": 741.917590527 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nAndrea M. Jokisaari, Khadija Mahbuba, Yuhao Wang, and Benjamin Beeler. The impact of anisotropic thermal expansion on the isothermal annealing of polycrystalline \u03b1-uranium. Computational Materials Science, 205:111217, Apr 2022. URL: https://doi.org/10.1016/j.commatsci.2022.111217, doi:10.1016/j.commatsci.2022.111217.\nHenry Proudhon, Nicolas Gu\u00e9ninchault, Samuel Forest, and Wolfgang Ludwig. Incipient bulk polycrystal plasticity observed by synchrotron in-situ topotomography. Materials, 11:2018, Oct 2018. URL: https://doi.org/10.3390/ma11102018, doi:10.3390/ma11102018.\nJuan Manuel Garc\u00eda and Thilo F. Morgeneyer. Strength and fatigue strength of a similar ti\u20106al\u20102sn\u20104zr\u20102mo\u20100.1si linear friction welded joint. Fatigue & Fracture of Engineering Materials & Structures, 42:1100-1117, Mar 2019. URL: https://doi.org/10.1111/ffe.12973, doi:10.1111/ffe.12973.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_t3X9GkiKh5YtIwvvFGFtqFju", + "info": { + "start_ts": 703.565379123, + "end_ts": 745.716529263 + } + }, + { + "role": "tool", + "content": "Retrieved Papers:\n## 1. A Roadmap for Edge Computing Enabled Automated Multidimensional Transmission Electron Microscopy\nValid Text Names:\n'mukherjee2022aroadmapfor pages 1-2', 'mukherjee2022aroadmapfor pages 2-3', 'mukherjee2022aroadmapfor pages 3-4', 'mukherjee2022aroadmapfor pages 4-6', 'mukherjee2022aroadmapfor pages 6-8', 'mukherjee2022aroadmapfor pages 8-9', 'mukherjee2022aroadmapfor pages 9-10', 'mukherjee2022aroadmapfor pages 10-11', 'mukherjee2022aroadmapfor pages 11-11'\n\nBibTex:\n@article{mukherjee2022aroadmapfor,\n author = \"Mukherjee, Debangshu and Roccapriore, Kevin M. and Al-Najjar, Anees and Ghosh, Ayana and Hinkle, Jacob D. and Lupini, Andrew R. and Vasudevan, Rama K. and Kalinin, Sergei V. and Ovchinnikova, Olga S. and Ziatdinov, Maxim A. and Rao, Nageswara S.\",\n title = \"A Roadmap for Edge Computing Enabled Automated Multidimensional Transmission Electron Microscopy\",\n year = \"2022\",\n journal = \"Microscopy Today\",\n volume = \"30\",\n pages = \"10-19\",\n month = \"Oct\",\n doi = \"10.48550/arxiv.2210.02538\",\n url = \"https://doi.org/10.48550/arxiv.2210.02538\",\n publisher = \"arXiv\"\n}\n\n\nAbstract:\nAbstract: The advent of modern, high-speed electron detectors has made the collection of multidimensional hyperspectral transmission electron microscopy datasets, such as 4D-STEM, a routine. However, many microscopists find such experiments daunting since analysis, collection, long-term storage, and networking of such datasets remain challenging. Some common issues are their large and unwieldy size that often are several gigabytes, non-standardized data analysis routines, and a lack of clarity about the computing and network resources needed to utilize the electron microscope. The existing computing and networking bottlenecks introduce significant penalties in each step of these experiments, and thus, real-time analysis-driven automated experimentation for multidimensional TEM is challenging. One solution is to integrate microscopy with edge computing, where moderately powerful computational hardware performs the preliminary analysis before handing off the heavier computation to high-performance computing (HPC) systems. Here we trace the roots of computation in modern electron microscopy, demonstrate deep learning experiments running on an edge system, and discuss the networking requirements for tying together microscopes, edge computers, and HPC systems.\n\nRelevant Snippet:\n\u2026 Elastically scattered electron diffraction patterns (the 4D-\u2026 allows flexible Python-based control of the microscope hardware \u2026 This has led to development of open-source microscope data \u2026\n\n\n---\n\n## 2. Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook.\nValid Text Names:\n'botifoll2022machinelearningin pages 1-2', 'botifoll2022machinelearningin pages 3-4', 'botifoll2022machinelearningin pages 4-6', 'botifoll2022machinelearningin pages 6-7', 'botifoll2022machinelearningin pages 7-8', 'botifoll2022machinelearningin pages 8-9', 'botifoll2022machinelearningin pages 9-10', 'botifoll2022machinelearningin pages 10-12', 'botifoll2022machinelearningin pages 13-14', 'botifoll2022machinelearningin pages 14-15', 'botifoll2022machinelearningin pages 17-18', 'botifoll2022machinelearningin pages 19-21', 'botifoll2022machinelearningin pages 21-23', 'botifoll2022machinelearningin pages 23-24', 'botifoll2022machinelearningin pages 26-27', 'botifoll2022machinelearningin pages 30-31', 'botifoll2022machinelearningin pages 31-32', 'botifoll2022machinelearningin pages 32-33', 'botifoll2022machinelearningin pages 33-35', 'botifoll2022machinelearningin pages 35-36', 'botifoll2022machinelearningin pages 40-41', 'botifoll2022machinelearningin pages 42-43', 'botifoll2022machinelearningin pages 45-46', 'botifoll2022machinelearningin pages 47-48', 'botifoll2022machinelearningin pages 50-51'\n\nBibTex:\n@article{botifoll2022machinelearningin,\n author = \"Botifoll, Marc and Pinto-Huguet, Ivan and Arbiol, J.\",\n title = \"Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook.\",\n year = \"2022\",\n journal = \"Nanoscale horizons\",\n month = \"Oct\",\n doi = \"10.1039/d2nh00377e\",\n url = \"https://doi.org/10.1039/d2nh00377e\"\n}\n\n\nAbstract:\nIn the last few years, electron microscopy has experienced a new methodological paradigm aimed to fix the bottlenecks and overcome the challenges of its analytical workflow. Machine learning and artificial intelligence are answering this call providing powerful resources towards automation, exploration, and development. In this review, we evaluate the state-of-the-art of machine learning applied to electron microscopy (and obliquely, to materials and nano-sciences). We start from the traditional imaging techniques to reach the newest higher-dimensionality ones, also covering the recent advances in spectroscopy and tomography. Additionally, the present review provides a practical guide for microscopists, and in general for material scientists, but not necessarily advanced machine learning practitioners, to straightforwardly apply the offered set of tools to their own research. To conclude, we explore the state-of-the-art of other disciplines with a broader experience in applying artificial intelligence methods to their research (e.g., high-energy physics, astronomy, Earth sciences, and even robotics, videogames, or marketing and finances), in order to narrow down the incoming future of electron microscopy, its challenges and outlook.\n\n\nUnobtainable Papers:\nGear\u00f3id Liam Mangan, Grigore Moldovan, and Andrew Stewart. Influence: an open-source python package to model images captured with direct electron detectors. Microscopy and microanalysis : the official journal of Microscopy Society of America, Microbeam Analysis Society, Microscopical Society of Canada, 29 4:1380-1401, Jul 2023. URL: https://doi.org/10.1093/micmic/ozad064, doi:10.1093/micmic/ozad064.\nBenjamin Miller, Anahita Pakzad, Benjamin Savitzky, Colin Ophus, and Cory Czarnik. Using py4dstem in gms: hybrid open-source, commercial-freeware methods for analyzing 4d stem datasets. Microscopy and Microanalysis, 27:1352-1354, Jul 2021. URL: https://doi.org/10.1017/s1431927621005043, doi:10.1017/s1431927621005043.\nRandi Holmestad, Elisabeth Thronsen, Yasuhito Kawahara, Tina Bergh, J\u00f8rgen A S\u00f8rhaug, Christoph M Hell, Ruben Bj\u00f8rge, Emil F Christiansen, Kenji Kaneko, and Calin D Marioara. Nucleation and phase development of precipitates in age-hardenable aluminium alloys studied by 4d-stem. Microscopy and microanalysis : the official journal of Microscopy Society of America, Microbeam Analysis Society, Microscopical Society of Canada, 29 Supplement_1:305-306, Jul 2023. URL: https://doi.org/10.1093/micmic/ozad067.141, doi:10.1093/micmic/ozad067.141.\nH. W. Leung, R. C. B. Copley, J. Laulainen, D. N. Johnstone, and P. A. Midgley. Combining scanning nanobeam electron diffraction with 3d electron diffraction to investigate crystal defects. Microscopy and Microanalysis, Jul 2024. URL: https://doi.org/10.1093/mam/ozae044.901, doi:10.1093/mam/ozae044.901.\nFernando C Castro and Anahita Pakzad. Optimizing 4d stem data collection and processing for crystal orientation mapping analysis in digitalmicrograph. Jul 2025. URL: https://doi.org/10.1093/mam/ozaf048.096, doi:10.1093/mam/ozaf048.096.\nGear\u00f3id Liam Mangan. Advancements and applications of (scanning) transmission electron microscopy simulations. Text, Jan 2022. URL: https://doi.org/10.34961/researchrepository-ul.21621174, doi:10.34961/researchrepository-ul.21621174.\nPavlina Sikorova, Miroslav Slouf, Tomas Molnar, and Vladislav Krzyzanek. Computational methods for automated center determination in electron diffraction patterns. Journal of Applied Crystallography, 59(3):845-857, Apr 2026. URL: https://doi.org/10.1107/s1600576726002384, doi:10.1107/s1600576726002384.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_2WFpaGkkorc7QNUjgoGtFJOi", + "info": { + "start_ts": 703.598195246, + "end_ts": 765.667408384 + } + }, + { + "role": "tool", + "content": "Search retrieved no additional papers. This may be because nothing matched the search query, or because all found papers have been previously retrieved.\n\nUnobtainable Papers:\nDeveloping a novel hierarchical approach for multiscale structural reliability predictions for ultra-high consequence applications\nINVESTIGATION OF SINGLE CRYSTAL AND BI-CRYSTAL DEFORMATION IN BODY-CENTERED CUBIC TANTALUM USING INDENTATION\nCaleb E. Andrews, Maria Strantza, Nicholas P. Calta, Manyalibo J. Matthews, and Mitra L. Taheri. A denoising autoencoder for improved kikuchi pattern quality and indexing in electron backscatter diffraction. Ultramicroscopy, 253:113810, Nov 2023. URL: https://doi.org/10.1016/j.ultramic.2023.113810, doi:10.1016/j.ultramic.2023.113810.\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: []\n\nStatus: Paper Count=22 | Relevant Papers=0 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=0 | Current Cost=$0.0481 | Disease-Target Evidence=0", + "name": "paper_search", + "tool_call_id": "call_xqX7V5hmrKT4tjzjK7lLSxlE", + "info": { + "start_ts": 703.631840635, + "end_ts": 739.262894677 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_QhQG5TaOdTjPfKLBDrG4HStR", + "type": "function", + "function": { + "arguments": "{\"question\": \"Extract evidence relevant to comparing EBSD post-processing tools: (1) MTEX grain reconstruction and texture/ODF capabilities; (2) DREAM.3D grain segmentation and 2D/3D EBSD processing; (3) use of KAM or grain boundaries in EBSD workflows; (4) peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, PyMicro. Include exact capabilities and limitations where stated.\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Most relevant evidence (with ids):\n\n## pqac-00000000\n\nFrom text 'folwarczny2026accurategrainboundary pages 5-9'.\n\nMartin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal.\n\nThe excerpt describes using MTEX and DREAM.3D for EBSD analysis. MTEX capabilities include: plotting inverse pole figure (IPF) maps and orientation distribution function (ODF) pole figures for texture analysis, grain size distribution analysis using equivalent circle radius function with lognormal fitting, grain shape analysis via ellipse fitting with aspect ratio and characteristic shape function calculations, and grain boundary segment orientation analysis. DREAM.3D is used for simulating 3D microstructures and exporting 2D planar sections in multiple orthogonal directions (XY, XZ, YZ) with proper reference frame rotations to match experimental EBSD conventions. The workflow converts DREAM.3D output to standard .ang format used in TSL OIM Analysis. No information is provided regarding KAM usage, orix, DefDAP, kikuchipy, pyxem, or PyMicro tools.\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000]\n\nStatus: Paper Count=22 | Relevant Papers=1 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=1 | Current Cost=$0.0899 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_QhQG5TaOdTjPfKLBDrG4HStR", + "info": { + "start_ts": 779.597149611, + "end_ts": 787.125253756 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_at4kevxNAZzrz4Vmrs3X4MTb", + "type": "function", + "function": { + "arguments": "{\"question\": \"What does the literature say specifically about DREAM.3D grain segmentation algorithms, misorientation thresholds, 2D/3D EBSD reconstruction, automation/pipelines, and maturity?\"}", + "name": "gather_evidence" + } + }, + { + "id": "call_1oiOo0bOuigk4ZUPXbdj7xZ2", + "type": "function", + "function": { + "arguments": "{\"text_name_list\": [\"polonsky2019threedimensionalanalysisand pages 1-2\", \"polonsky2019threedimensionalanalysisand pages 2-5\", \"stanek2020comparisonofsegmentation pages 1-2\", \"stanek2020comparisonofsegmentation pages 2-7\", \"savage2021anautomatedprocedure pages 1-2\", \"wanni2024machinelearningenhanced pages 1-2\"]}", + "name": "read_text" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Most relevant evidence (with ids):\n\n## pqac-00000001\n\nFrom text 'polonsky2019threedimensionalanalysisand pages 1-2'.\n\nAndrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\nThe excerpt discusses DREAM.3D as a mature software package for 3D EBSD data reconstruction and microstructure analysis that aggregates community-developed algorithms. It operates as a multi-step pipeline with numerous input variables and outputs. Traditionally, DREAM.3D runs serially on workstations with predefined parameters. The authors integrated DREAM.3D into a cloud-based platform (BisQue) to enable parallelization and parameterization of pipeline variables across value ranges. This parallelization specifically provided insights about grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets of additively manufactured materials. The manual, serial approach previously limited exploration of the full parameter space due to long computation times scaling with dataset size and complexity, introducing bias in reconstruction processes.\n\n\n---\n\n## pqac-00000002\n\nFrom text 'polonsky2019threedimensionalanalysisand pages 13-14'.\n\nAndrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\nThe excerpt discusses DREAM.3D's integration into the BisQue cloud platform for 3D microstructure reconstruction. Key findings include: (1) grain segmentation misorientation tolerance optimization specifically for additively manufactured materials yielded new insights; (2) algorithms for 2D and 3D volume element sampling revealed that traditional 2D metallographic sectioning produces grain size measurements approximately half those of equivalent 3D measurements in heterogeneous AM microstructures; (3) the platform enables pipeline parameterization and parallelized module execution for reconstruction optimization; (4) automation includes massively parallelized slicing/cropping pipelines to simulate 2D versus 3D statistical grain measurements. The work demonstrates DREAM.3D's practical application for addressing discrepancies between 2D and 3D grain analysis in additively manufactured materials.\n\n\n---\n\n## pqac-00000003\n\nFrom text 'polonsky2019threedimensionalanalysisand pages 2-5'.\n\nAndrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\nThe excerpt describes DREAM.3D's application for 3D EBSD reconstruction and grain segmentation in additively manufactured materials. Grains are segmented using misorientation tolerance thresholds, with specific implementation details provided for 304L stainless steel (misorientation tolerance with minimum grain size of 64 voxels) and another alloy (2\u00b0 misorientation tolerance with 125 voxels minimum). The tool supports 2D EBSD slice reconstruction into coherent 3D volumes. Key automation features include parameterizable pipeline execution allowing parallel processing across multiple compute nodes with variable parameter ranges rather than discrete values. This parallelization capability reduces setup and analysis time for parameter studies. The excerpt indicates DREAM.3D was integrated with BisQue cloud infrastructure, suggesting mature implementation for workflow automation and data provenance tracking, though the excerpt does not extensively discuss algorithm maturity or theoretical foundations.\n\n\n---\n\n## pqac-00000004\n\nFrom text 'yi2025rapidgrainsegmentation pages 4-7'.\n\nYu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal.\n\nThe excerpt describes Dream3D's application to 2D/3D EBSD reconstruction from the Small IN100 dataset (117 slices of EBSD data with 189 x 201 scan points). Dream3D employs automated pipelines for grain boundary reconstruction, including full reconstruction, morphological statistics, crystallographic statistics, mesh generation, and grain boundary character distribution (GBCD) analysis. The segmentation achieved high performance metrics (average IoU of 91.58% and Dice score of 95.6%). However, limitations exist: the algorithm relies on Inverse Pole Figure (IPF) color contrast and can misidentify neighboring grains with similar IPF colors as the same grain, causing missegmentation. The excerpt acknowledges this as an inherent uncertainty when using this pipeline for grain segmentation. No specific information about misorientation thresholds is provided. The text notes that EBSD remains necessary when precise grain orientation information is required, suggesting Dream3D is best suited for rapidly quantifying grain size distribution with approximately 10% error tolerance.\n\n\n---\n\n## pqac-00000005\n\nFrom text 'stanek2020comparisonofsegmentation pages 1-2'.\n\nJ. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\nThe excerpt mentions DREAM.3D software as a computational tool that enables significant data evaluation and five-parameter grain boundary analysis, though it does not elaborate on specific segmentation algorithms. Regarding misorientation thresholds, the paper notes that typical values are 5\u00b0 in EDAX OIM software, with other choices possible. The work demonstrates critical differences between 2D and 3D EBSD segmentation: identical misorientation thresholds produce different results in 2D versus 3D cases because nonconvex grains may appear as separate grains in 2D slices but belong to single grains in 3D reconstruction. The paper emphasizes that proper segmentation attention is necessary to avoid misleading quantitative microstructure results. However, the excerpt provides limited information about automation/pipelines or the overall maturity level of these technologies.\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005]\n\nStatus: Paper Count=22 | Relevant Papers=4 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=6 | Current Cost=$0.1324 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_at4kevxNAZzrz4Vmrs3X4MTb", + "info": { + "start_ts": 805.10017757, + "end_ts": 814.30554037 + } + }, + { + "role": "tool", + "content": "## 1. wanni2024machinelearningenhanced pages 1-2 (Context ID: pqac-00000006)\n\n npj | computational materials Article\n\n Published in partnership with the Shanghai Institute of Ceramics of the Chinese Academy of Sciences\n\n\n\n\n\n https://doi.org/10.1038/s41524-024-01324-4\n Machine learning enhanced analysis of\n EBSD data for texture representation\n\n\n\n Check for updates\n\n J. Wanni 1, C. A. Bronkhorst 1,2 & D. J. Thoma 1,2\n\n\n\n Generating reduced-order, synthetic grain structure datasets that accurately represent the measured\n grain structure of a material is important for reducing the cost and increasing the accuracy of\n computational crystal plasticity efforts. This study introduces a machine-learning-based approach,\n termed texture adaptive clustering and sampling (TACS), for generating representative Euler angle\n datasets that accurately mimic the crystallographic texture. The TACS approach employs K-means\n clustering and density-based sampling in a closed-loop iteration to create representative Euler angle1234567890():,; 1234567890():,;\n datasets. Proof-of-principle experiments were performed on rolled and recrystallized low-carbon\n steel. Validation of the TACS approach was extended to twenty-two datasets, varying lattice\n structures, and complex crystallographic textures, thereby encompassing a broad range of materials\n and crystal structures. Kolmogorov-Smirnov (K-S) test comparisons underscore the performance of\n the TACS approach over traditional electron backscatter diffraction EBSD dataset reduction\n techniques, with average K-S test scores nearing 0.9, indicating a high-\ufb01delity representation of the\n original datasets. In contrast, conventional methods display scores below 0.3, indicating less reliability\n of the structure representation. The independence of the TACS approach from material texture and its\n capability to autonomously generate datasets with predetermined data points demonstrates its\n unbiased potential in streamlining dataset preparation for crystallographic analysis.\n\n\n\n The microstructure of polycrystalline metallic materials is a very simulations are done by synthesizing a grain structure that mimics the\n important factor in determining their physical properties1\u20134. Pre- microstructural characteristics of the material. Typically, the information\n dominantly, the grain structure stands as the most observable micro- for these grain structures is initially obtained through electron backscatter\n structural feature in most polycrystalline metals, wielding a signi\ufb01cant diffraction (EBSD) datasets, which provide detailed mappings of grain\n in\ufb02uence on the structural behavior of the metal1,5,6. Speci\ufb01cally, aspects orientations. A great deal of effort is required to ensure that the character-\n such as grain shape (morphological texture) and lattice orientation istics of the synthesized grain structure accurately re\ufb02ect the actual micro-\n (crystallographic texture) can lead to mechanical property directional structure. Grain morphology is usually modeled using 3D software such as\n anisotropy5\u20138. In addition, morphological and crystallographic texture Dream3D, where the number of grains in the synthetic structure can vary\n can also in\ufb02uence the contribution of different deformation mechan- from a few tens to thousands, in\ufb02uenced by the available computational\n isms, such as dislocation slip and deformation twinning, to the overall resources and the choice of the crystal plasticity simulation approach13\u201316.\n deformation behavior of a metallic material7,8. As such, the grain However, while typical EBSD datasets for grain structures may\n structure has the potential to be engineered for enhancing speci\ufb01c encompass hundreds to thousands of grains, the grains included in a\n properties like strength and ductility7,9\u201311. Accurately de\ufb01ning the cor- representative volume element (RVE) might be reduced to manage\n relation between the grain structure of a metal and its properties is computational expenses effectively. The extent of this reduction can\n critical for developing useful computational methods for predicting and vary, depending on the speci\ufb01c requirements and objectives of the\n designing the mechanical response of materials. simulation. Such reductions necessitate meticulous attention to ensure\n Crystal plasticity simulations offer a powerful tool for studying the that the comprehensive grain information is accurately represented in\n in\ufb02uence of grain morphology and lattice orientation on the mechanical the RVE13,15\u201317,19. Given the complexities involved in capturing such\n response of crystalline materials12\u201318. Together with a theory representing detailed microstructural information within a reduced-order dataset,\n the elastic-plastic response of the single crystals, crystal mechanics maintaining the integrity of the crystallographic texture without\n\n\n\n 1Department of Materials Science and Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA. 2Department of Mechanical Engineering, Uni-\n versity of Wisconsin-Madison, Madison, WI 53706, USA. e-mail: dthoma@wisc.edu\n\n\n\n npj Computational Materials | (2024)10:133 1https://doi.org/10.1038/s41524-024-01324-4 Article\n\noverwhelming computational resources is a challenge in crystal plasti- Results and discussion\ncity simulations13,15\u201317,19. Proof-of-principle evaluation\n Generating accurate reduced-order Euler angle datasets is rele- The pole \ufb01gure (PF) maps of the EBSD dataset of a rolled and recrystallized\nvant not only for crystal plasticity simulations but also for enhancing low carbon (0.23 wt.% C) steel that was used to develop the algorithm are\nthe effectiveness of other modeling techniques such as reduced-order depicted in Fig. 1a.The rolling direction (RD) and normal direction (ND) of\nmodeling and surrogate modeling20\u201326. These methods aim to the 12.7 mm thick plate are indicated for reference. The EBSD analysis\nstreamline complex physical models, facilitating faster simulations identi\ufb01ed onlytheiron-BCCphase,resultingfromthequenchingstepinthe\nwhile maintaining high accuracy. For instance, in reduced-order manufacturing process. There is a possibility that EBSD did not detect\nmodeling, key characteristics of a material\u2019s behavior can be ef\ufb01- retained austenite or carbides in the material or mis-indexed the patterns39.\nciently encapsulated with fewer degrees of freedom, proving essential However,thisisnotrelated to thescopeof thisstudy.Astep sizeof 0.250 \u03bcm\nfor real-time simulations and iterative design processes21\u201324,26. Simi- was used to collect the data and the dataset consists of 165088 indexed\nlarly, surrogate models, which act as proxies for more intricate points. A roughly 120 \u00d7 90 \u03bcm2 region was scanned. The dataset consists of\nsimulations, can be optimized through the use of reduced-order 2068 grains, hence the dataset is large enough to represent the grain\ndatasets that capture microstructural characteristics critical to structure of the material. The pole \ufb01gure (PF) maps reveal a weak (100)\nmaterial properties20,25. As these modeling approaches are still evol- texture\u2014twice the random distribution\u2014in the rolling direction of the\nving, particularly in their capacity to predict material deformation plates. The (110) and (111) directions also show weak textures at an angle of\nbehavior, they stand to gain signi\ufb01cantly from advanced methods that 45 degrees to the RD, possibly due to the (100) texture in the RD. The weak\nef\ufb01ciently compress dataset sizes while accurately representing the texture of the EBSD dataset, characterized by more randomly oriented\nmicrostructure of the material. grains with a broad variation in orientations, poses a signi\ufb01cant challenge in\n Machine learning (ML) based approacheshave already been utilized to generating a representative reduced-order synthetic Euler angle dataset.\nsolve complex problems that are associated with EBSD datasets27\u201330. For Accurately capturing the diverse orientation characteristics of such weakly\ninstance, K. Kaufman et al. demonstrated the use of deep learning, speci- textured materials typically requires a large number of g\n\n## 2. stanek2020comparisonofsegmentation pages 2-7 (Context ID: pqac-00000007)\n\nture. by a boundary de\ufb01ned by a given threshold of misori-\nThe polycrystalline microstructures are observed by entation so that the sets are detected as two grains,\n3D EBSD and 2D EBSD. The approaches presented while in 3D case, all these points belong to one larger\nin Section 2 for simulated toy data examples are dis- set which is surrounded by a boundary. Therefore in\ncussed subsequently for both materials. The curve of the 3D case, all these points lie in the same grain,\ndependence of the mean number of grains, in a single see Fig. 1. This behaviour can be observed when the\n2D slice, on the misorientation tolerance level, is plot- grains are nonconvex, or when there is a boundary of\nted for both 2D and 3D methods. The di\ufb00erences are the given misorientation inside the grain. One could\nremarkable. conclude such a situation to be unrealistic, but the\n In Section 4, as an application to microstructure same behaviour can be observed in materials formed\nanalysis of 3D EBSD data, we present a \ufb01ve-parameter by convex grains with very low local misorientation\nanalysis of grain boundaries in Cu specimen and sug- (all voxels from a single grain have nearly the same\ngest a physical interpretation. Three levels of seg- orientation), if the chosen misorientation threshold is\nmentation are considered and discussed. The 3D mi- higher than the minimal misorientation of the neigh- J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 303\n\n\n\n\n\nFig. 2. A Laguerre tessellation simulated from the 3D\nstochastic Gibbs-Laguerre model in a cubic window. The\npattern (a) involves 1560 grains coloured at random. In\n(b), there is a 2D slice taken from the middle of the pat-\n tern (a). Fig. 3. Histograms of misorientations between neighbour-\n ing grains in the simulated data. (a) the case (A) \u2013 low\n angle grain boundaries, (b) the case (B) \u2013 high angle grain\n boundaries.\nbouring grains. In this case, the segmentation joins\nmore real grains into one larger grain. The situation\nis visualized in Fig. 1b. In both cases in the \ufb01gure, we tations of cubic crystal grid of grains were simulated,\nhave three grains which are segmented in 3D into one namely:\ngrain, because some boundaries (dashed lines) have (A) the misorientation between neighbouring\nlower misorientation than the chosen threshold, but grains is low, (see Fig. 3a),\nthe segmentation in 2D (in the cutting plane) sepa- (B) the misorientation between neighbouring\nrates these grains because the boundary in the cutting grains is high, (see Fig. 3b).\nplane (marked by the solid line) has greater misorien- These two cases represent the situations, when the\ntation than the chosen threshold. material prefers low angle grain boundaries (LAGB)\n In this section, we study the di\ufb00erence between or high angle grain boundaries (HAGB), respectively.\nsegmentation into grains in 2D and 3D for simulated For further purposes, we work with voxelized data\ndata. As a model for a polycrystalline microstructure, (100 \u00d7 100 \u00d7 100 voxels), where each voxel has its\nwe use the Laguerre tessellation [15], see Fig. 2. The orientation given by the associated grain. The data\nMarkov chain Monte Carlo method of simulation of a are processed by the software DREAM.3D. Since the\nGibbs-Laguerre tessellation [16] is extended by adding cells in the Laguerre tessellation (representing grains\nthe Euler angles to the state space. For this model of a in our case) are convex, and all voxels in each grain\ntessellation structure, two di\ufb00erent scenarios of orien- are assigned the same orientation, the simulated data304 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n\n\n\nFig. 4: The segmentation of simulated data for the case (A) in 2D with the misorientation threshold (a) 2\u25e6, (b) 2.5\u25e6, and\n(c) 3\u25e6and the same section from 3D reconstruction with the misorientation threshold (d) 2\u25e6, (e) 2.5\u25e6, and (f) 3\u25e6. The\n grains are coloured at random.\n\n\nrepresent a material without noise, i.e., all the grains into grains are considered, using the same \ufb01lters and\nare well de\ufb01ned. the same settings as in 3D. In this way, two di\ufb00erent\n To compare the results obtained in 2D and 3D, segmentations of the same slice are received, which are\nthe \ufb01lter \u2018Segment Features (Misorientation)\u2019 with a comparable. The \ufb01rst one represents the processing in\nchosen misorientation threshold for the entire 3D data 3D, while the second one represents the processing in\nset is used \ufb01rst. Then, a single slice from the processed 2D.\n3D data is cut out. This slice is treated separately, i.e., Just note that in the present study, the slice num-\nonly the voxels from the slice for the segmentation ber 50 is processed, which is shown in Fig. 2b. J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 305\n\n\n\n\n\nFig. 5. The segmentation of simulated data for the case (B) in 2D with the misorientation threshold (a) 2\u25e6, (b) 36\u25e6, and\n(c) 40\u25e6and the same section from 3D reconstruction with the misorientation threshold (d) 2\u25e6, (e) 36\u25e6, and (f) 40\u25e6. The\n grains are coloured at random.\n\n\n The di\ufb00erence between segmentation in 2D and 3D For the case (B), a similar result is shown in Fig. 5.\nin the case (A) is shown in Fig. 4. In this case, the mis- In this case, we use the misorientation thresholds equal\norientation thresholds equal to 2\u25e6, 2.5\u25e6, and 3\u25e6are used. to 2\u25e6, 35\u25e6, and 40\u25e6. Note that 0.01, 5.0, and 9.4 %\nNote that 0.15, 13.2, and 26.5 % of boundaries have of boundaries have misorientation lower than 2\u25e6, 35\u25e6,\nmisorientation lower than 2\u25e6, 2.5\u25e6, and 3\u25e6, respecti- and 40\u25e6, respectively. While we get similar segmenta-\nvely. While for the threshold of 2\u25e6, we get nearly the tions for all chosen thresholds between 2\u25e6and 35\u25e6, the\nsame segmentation in both cases, for the thresholds of di\ufb00erence between segmentations for the threshold of\n2.5\u25e6and 3\u25e6, the segmentations are entirely di\ufb00erent. 40\u25e6is not marginal.306 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n\n\n\nFig. 6. The segmentation of simulated data with the chosen misorientation threshold 2.5\u25e6for (a) 2D data, (b) 3D data \u2013\n10 slices, (c) 3D data \u2013 20 slices, (d) 3D data \u2013 30 slices, (e) 3D data \u2013 50 slices, and (f) 3D data \u2013 100 slices. The grains\n are coloured at random.\n\n\n The question is, how does the di\ufb00erence between number 50, which has already been studied. Six sce-\nsegmentation in 2D and 3D depend on the thickness of narios were compared: (a) segmentation in 2D, where\nthe specimen. Obviously, if we work with a 3D data set only the slice number 50 was used, (b) segmentation\nwhich contains only a few slices, the di\ufb00erence between in 3D with the data set of 10 slices enumerated 45\u201354,\nsegmentation in 2D and 3D should be marginal, since (c) segmentation in 3D with the data set of 20 slices\nthere is not enough space for the e\ufb00ect described in enumerated 40\u201359, (d) segmentation in 3D with the\nFig. 1. To study this dependence, the simulated data data set of 30 slices enumerated 35\u201364, (e) segmen-\n(A) were used, and the focus was given on the slice tation in 3D with the data set of 50 slices enumer- J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319 307\n\n\n tion in 2D and 3D for the cases (A), (C), and (D)\n is investigated. For all three cases, the segmentation\n with three misorientation thresholds is studied. The\n \ufb01rst one is 0.1\u25e6(called 0 % misorientation segmen-\n tation in the sequel), the second one is chosen so\n that 10 % of boundaries have lower misorientation\n (called 10 % misorientation segmentation in the se-\n quel). The last one corresponds to 15 % of boundaries\n with misorientation lo\n\n## 3. stanek2020comparisonofsegmentation pages 1-2 (Context ID: pqac-00000008)\n\n Kovove Mater. 58 2020 301\u2013319 301\n DOI: 10.4149/km 2020 5 301\n\n\n\n Comparison of segmentation of 2D and 3D EBSD measurements\n in polycrystalline materials\n\n\n J. Stan\u02c7ek1*, J. Kope\u010dek2, P. Kr\u00e1l3, I. Kara\ufb01\u00e1tov\u00e14, F. Seitl4, V. Bene\u01614\n\n\n\n 1Department of Mathematics Education, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n 2Institute of Physics of the CAS, Department of Functional Materials, Na Slovance 2, 182 00 Prague, Czech Republic\n 3Institute of Physics of Materials, ASCR, \u02c7Zi\u017ekova 22, 616 62 Brno, Czech Republic\n 4Department of Probability and Mathematical Statistics, Faculty of Mathematics and Physics, Charles University,\n 186 75 Prague, Czech Republic\n\n\n Received 30 December 2019, received in revised form 26 April 2020, accepted 17 June 2020\n\n\n Abstract\n\n A comparison study of grain segmentation in 2D and 3D electron backscatter di\ufb00rac-\n tion (EBSD) measurements of polycrystalline materials, based on a misorientation threshold,\n is presented. Using Laguerre tessellations with orientation marks, a simulation study that\n demonstrates a di\ufb00erent sensitivity of segmentation in the case of low angle and high angle\n grain boundaries is involved. Then two metallic materials measured by 3D EBSD are investi-\n gated concerning segmentation, and there are given some recommendations for the misorien-\n tation thresholds. Computational tools such as DREAM 3D software enable signi\ufb01cant data\n evaluation that is completed by a \ufb01ve-parameter analysis of grain boundaries.\n\n\n K e y w o r d s: electron backscatter di\ufb00raction (EBSD), grain boundaries, misorientation,\n tessellation\n\n\n\n 1. Introduction observed the boundaries \u2013 the separators, whereas\n grains or rather their interiors were homogeneous ar-\n The electron backscattered di\ufb00raction method eas inside with de\ufb01ned or expected value of properties.\n(EBSD) has been a powerful tool in material science Of course, it is known that a grain structure with the\nsince the nineties, when a certain level of computeri- tessellation formula can be created [7]; nevertheless,\nzation and application of Hough transform took place, EBSD is the \ufb01nal step reversing axiom to construct\nsee [1, 2]. Nowadays, we can perform not only 2D the grains [2]. We have a set of measured points in the\nEBSD but exploiting focused ion beams (FIB), even EBSD method. We have positions of such points and\n3D EBSD, by materials\u2019 slicing tomography [3]. The their marks (e.g., Euler angles). Thus, here the inte-\n3D EBSD data evaluation is a complicated process [4]. rior is a quality, which exists, whereas the boundary\nStill, we expect to be able to construct grains [5] and has to be constructed. The easiest way how to create\nmicrostructure parameters by applying the same or grains from the set of measured points is the following\nsimilar procedures as in 2D EBSD data evaluation [6]. de\ufb01nition: the grain interior shall have the same or\nHowever, a discrepancy in the segmentation of data very similar properties. Thus, we construct grains by\npoints into the grains in the 2D and 3D cases is ap- collecting the measured points with the de\ufb01ned prop-\nparent, since a nonconvex grain in 3D can be observed erty. The metallographic and EBSD grains must be\nas more than one grain in a 2D slice. then the same in the optimal case. However, the re-\n The segmentation problem is deeply connected to sult of an experiment is not optimal. It contains some\nthe basic understanding of the polycrystals. Gener- noise or even a kind of systematic aberrations, and the\nally, the de\ufb01nition of grains as they were observed by reconstruction of grain structure can be done only up\nfounders of metallography was purely natural: They to some level of accuracy.\n\n*Corresponding author: tel.: +420 951553215; e-mail address: stanekj@karlin.m.cuni.cz302 J. Stan\u02c7ek et al. / Kovove Mater. 58 2020 301\u2013319\n\n\n The problem described above was omitted in com-\nputational materials science since the 3D data is fre-\nquently used in an elementary manner. Despite the\nexplosion of three-dimensional methods in the last\ndecades, the majority of 3D models use simple math-\nematical schemes to build the structure by a general\nprocedure approaching the reality by just a few pa-\nrameters, measured from 2D cuts of material [8].\n Data obtained by 3D EBSD or 2D EBSD form a\nset of orientations of measured points in a 3D ma-\ntrix (the points are called voxels in the sequel) or\n2D matrix (pixels in the sequel), respectively. The\nEBSD method considers a grain to be the set of all\npoints (voxels or pixels, respectively) with a similar\norientation, which are surrounded by a boundary of\na given misorientation threshold. The typical thresh-\nold in the EDAX OIM software is 5\u25e6[9], but other\nchoices are possible as well [10, 11]. The obtained seg-\nmentation into the grains depends on the chosen mis- Fig. 1. Di\ufb00erences between slices processed from 2D EBSD\n and the same slice processed from 3D EBSD. The greyorientation threshold, of course, but even if we use a\n colour displays the real grains (in the plane, which is or-\n\ufb01xed misorientation threshold, we may get di\ufb00erent\n thogonal to the cutting plane). The blue colour displays\nresults of the segmentation in 2D and 3D, as shown their 3D segmentation in the cutting plane, i.e., the parts\nin [12]. Therefore, the results obtained in 2D EBSD are considered to belong to the same grain. Finally, the\nmay not be easily compared with the results from 3D red and the orange colours show their 2D segmentation\nEBSD. in which they are considered to be di\ufb00erent grains. (a)\n In the paper, we present a comparison study of The situation with a nonconvex grain or a grain with a\nsegmentation of 2D and 3D EBSD measurements in boundary inside. (b) The situation with convex grains,\n boundaries with the misorientation smaller than the cho-polycrystalline materials, based on a misorientation\n sen threshold are denoted by the dashed line.\nthreshold. The main results of this topic are presented\nin Sections 2 and 3. The aim is to draw the reader\u2019s\nattention to the danger of obtaining misleading quan-\ntitative results about the geometry of microstructure crostructure analyses make use of DREAM 3D soft-\nwhen paying little attention to a proper segmentation. ware [14].\nFirst, in Section 2, the problem is demonstrated in a Some conclusions from Sections 2 and 3 are given\nsimulation study. A generalized Laguerre tessellation in more detail in the \ufb01nal part of the paper.\nmodeling of polycrystalline microstructure measured\nby 3D EBSD was used in [13]. We can simulate this\nmodel with a priori given distribution of misorienta- 2. Di\ufb00erences between 2D and 3D\ntions. The in\ufb02uence of this distribution on the seg- segmentation\nmentation is shown.\n In Section 3, we discuss the segmentation prob- The di\ufb00erences between segmentation into grains\nlem with two materials, namely the AlMgSc alloy af- in 2D and 3D are caused by the fact that in a 2D\nter ECAP and annealing, and pure copper after high- cutting plane, two sets of voxels may be surrounded\npressure torsion and annealing at room temperature. by a boundary de\ufb01ned by a given threshold of misori-\nThe polycrystalline microstructures are observed by entation so that the sets are detected as two grains,\n3D EBSD and 2D EBSD. The approaches presented while in 3D case, all these points belong to one larger\nin Section 2 for simulated toy data examples are dis- set which is surrounded by a boundary. Therefore in\ncussed subsequently for both materials. The curve of the 3D case, all these points lie in the same grain,\ndependence of the mean number of grains, in a single see Fig. 1. This behaviour can be observed when the\n2D slice, on the misorientation tolerance level, is plot- grains are nonconvex, or when there is a boundary of\nted for both 2D and 3D meth\n\n## 4. savage2021anautomatedprocedure pages 1-2 (Context ID: pqac-00000009)\n\n Materials Characterization 171 (2021) 110808\n\n Contents lists available at ScienceDirect\n Materials Characterization\n\n\n journal homepage: www.elsevier.com/locate/matchar\n\n\nAn automated procedure built on MTEX for reconstructing deformation\ntwin hierarchies from electron backscattered diffraction datasets of heavily\ntwinned microstructures\nDaniel J. Savage a,b, Rodney J. McCabe b, Marko Knezevic a,*\na Department of Mechanical Engineering, University of New Hampshire, Durham, NH 03824, USA\nb Materials Science and Technology Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA\n\n\nA R T I C L E I N F O A B S T R A C T\n\nKeywords: This paper presents a set of algorithms built on the MTEX and MATLAB graph toolboxes for automatic recon\u00ad\nMicrostructures struction of deformation twin hierarchies from Electron Backscatter Diffraction (EBSD) datasets with a focus on\nTwinning developing methods for heavily twinned microstructures (twin fractions >0.5). The algorithms address key issues\nTitanium arising at large strains, mainly: missing twin relationships, grouping of heavily deformed grain fragments intoEBSD families of similar orientation originating from a single initial grain, identification of parent fragments for largeMTEX\n twin volume fractions, and classification of families having twin relationships with multiple families. To facilitate\n the development of these algorithms, large-grained ultra-high purity \u03b1-Ti deformed in compression along two\n directions is investigated. Graphs are utilized to handle non-local geometric merging and to represent re\u00ad\n lationships throughout the reconstruction process. When determining if a grain fragment is from the undeformed\n microstructure, the combined metrics of the fragment\u2019s orientation volume fraction in the initial texture and the\n directed graph centrality measure of out-closeness (the number of nodes reached in a graph from a given node)\n are essential. To address automation in reconstructing the sequence of twinning and relating fragments origi\u00ad\n nating from a single grain in the initial microstructure, the twin family tree is formulated as a minimum spanning\n tree emanating from the initial grain family. A scheme constructing the distances associated with twin rela\u00ad\n tionship comprising the spanning tree is developed, and a novel quasi-directional Prim spanning tree algorithm is\n used to determine the twin family tree. The procedure is demonstrated to significantly improve the level of\n automation in reconstructing twin hierarchies in heavily twinned microstructure compared to other methodol\u00ad\n ogies in literature. The procedure can readily be applied to analyses of twinning in metals, as well as provide an\n approach for routinely extracting twin statistics at larger deformation levels than previously possible. Signifi\u00ad\n cantly, the procedure is demonstrated to be capable of identifying third generation twinning in \u03b1-Ti\n microstructures.\n\n\n1. Introduction formation in the vicinity of their boundaries [9\u201315]. This correlation has\n been rationalized to arise due to substantial crystal reorientations and\n Deformation twinning is a shear-based transformation that results in underlying shifts in active crystallographic slip modes within the twin\u00ad\ncrystal reorientation and significantly influences the evolution of local ned domain. Furthermore, collection of nucleation statistics such as in\nand overall material properties of many important metals undergoing fatigue loading, is vital to building the understanding of fracture in\ntwinning [1\u20133]. One well-known twin phenomena is the orientation hexagonal close-packed (HCP) materials [16]. Thus, measuring and\ndependent anisotropic hardening introduced by twin lamella frag\u00ad classifying twinning is relevant for developing a working knowledge of\nmenting slip domains and reorienting towards hard crystal orientations fracture and the anisotropic response of materials.\n[4\u20138]. In addition to hardening, many studies show the formation of Twinning plays an important role in developing Integrated Compu\u00ad\ntwins are correlated to flow localization and ultimately void/crack tational Materials Engineering (ICME) toolsets. Crystal plasticity\n\n * Corresponding author at: University of New Hampshire, Department of Mechanical Engineering, 33 Academic Way, Kingsbury Hall, W119, Durham, NH 03824,\nUSA.\n E-mail address: marko.knezevic@unh.edu (M. Knezevic).\nhttps://doi.org/10.1016/j.matchar.2020.110808\nReceived 3 September 2020; Received in revised form 17 November 2020; Accepted 24 November 2020\nAvailable online 4 December 2020\n1044-5803/\u00a9 2020 Elsevier Inc. All rights reserved.D.J. Savage et al. Materials Characterization 171 (2021) 110808\n\n\n\n\n\nFig. 1. Schematic of the tasks and challenges performed in twin codes (from left to right): identification of twin relationships and grouping of grain fragments\n(clusters), identification of fragments of the same orientation (families, F), and the determination of the representative family tree. Note that a cluster is the group of\nfamilies comprising a single grain in the initial microstructure. Shown is an example of the family tree where the F1 family is the parent of families F2 and F4, and F3\nis a second generation twin whose parent is F2. A circular twin relationship (e.g. F1, F2, F3, and F1) arises from the combination of twin crystallographic reor\u00ad\nientations and a twin code must identify the correct twin hierarchy. The circular relationship causes no root (the family that is not a twin) to be evident based on the\ndirections of twin relationships, and another criterion (presented in this work) is needed to determine the root automatically.\n\nconstitutive laws, for instance, enable ICME [17\u201325] by modeling strain root by area majority rule, and relying heavily on fragment geometry to\npath dependent deformation behavior of highly anisotropic metals determine the twin. To address fundamental issues regarding post-\nthrough the evolution of slip, twin, and phase-induced texture evolution mortem analysis and slip, inverse modeling of the twinning process (e.\n[26,27]. In HCP and lower symmetry metals such as orthorhombic g. through physically based crystal plasticity modeling) in addition to\nmetals [28\u201333], it is usually necessary to emulate the twinning process, traditional twin identification approaches are needed. Key to this com\u00ad\nreproducing the statistics of experimental data to capture the mechan\u00ad bined approach is a highly flexible framework that has the capability to\nical response during complex loading [27,34,35]. The emulation of the assimilate all levels of microstructure and modeling information into a\ntwinning process requires detailed knowledge of the twin types, twin coherent interpretation of the microstructure evolution.\nfractions, twin nucleation criterion, and their role in hardening the Volume fractions of twins in highly deformed microstructures have\nmaterial; and ICME relies on the ability to extract twin-based been determined using boundary length fractions and texture based\ninformation. methods. The boundary length fraction method is based on measuring\n Considering the importance of twinning in fracture and under\u00ad the fraction of high-angle boundaries in the microstructure consistent\nstanding the anisotropic response of materials, several automatic twin with the twin misorientation relationship. The boundary fraction\nidentification\n\n## 5. polonsky2019threedimensionalanalysisand pages 2-5 (Context ID: pqac-00000010)\n\nAM.3D [8] plugin was\ntity, and resolution sufficient to make such measurements developed for BisQue in order to incorporate the materials\n[11]. Even in 2D EBSD analysis, such standards are rare science based 3D dataset analysis tools. The database\nand do not necessarily address additional subtleties arising architecture of BisQue is such that every processing\nfrom more complex, albeit commonly studied, microstruc- execution, dataset (input and output), and workflow\ntures, including twinned materials, nanocrystalline mate- processing pipeline receives a unique web address URL.\nrials, and deformed materials [12\u201317]. Standardization of These URLs are useful for at least two reasons: tracking\ncharacterization methods becomes critical as the complex- data provenance and data sharing. Data provenance with\nity of the analyses increases. For example, high-resolution BisQue is automatically captured and can be easily accessed\nEBSD (HR-EBSD) and Heaviside Digital Image Corre- through visualized workflows, as shown in Fig. 1. The\nlation (HDIC) [18] are extensions of conventional EBSD workflow shows where input data is located, the parameters\nand digital image correlation (DIC) that require specialized of a module execution are captured in the metadata, and\npost-processing to measure elastic and plastic strain at a the output data location is tracked. Therefore, passing\nresolution of 10-4 or smaller [19, 20] and 1.5\u00d710\u22123 [21], data between research colleagues becomes straightforward\nrespectively. The variety of analytic approaches available to due to the data operation transparency. Furthermore, data\nmake accurate strain measurements, as well as the sensi- sharing is facilitated due to the data all being stored\ntivity of these measurements to instrumentation and sample and accessed remotely, mitigating versioning problems\npreparation, motivates more rigorous standardization of the common to large data operations. For example, two\ntechnique via sample reference standards [22, 23]. groups collaborating between the US and Europe may findIntegr Mater Manuf Innov (2019) 8:37\u201351 39\n\n\n\n\n\nFig. 1 BisQue data provenance tree from an additively manufactured 304L Stainless Steel 3D dataset. Arrows show workflow between input data,\nDREAM.3D pipelines, module executions (mex), and associated datasets\n\n\n\ndifficulty operating on the same dataset version, especially In addition, a volume viewer enables web-based visual-\nwhen business hours do not overlap, without extensive ization of cell-level data, removing the need for programs\ncommunication. like HDFView or ParaView and their associated hardware\n requirements [31, 32]. 3D volumes collected via TriBeam\nBisQue DREAM.3D Integration tomography and reconstructed in DREAM.3D generally\n require many tens of gigabytes of RAM for visualiza-\nA DREAM.3D module was incorporated into BisQue to tion in ParaView, limiting analysis to powerful workstation\nallow for the processing and treatment of 3D datasets computers.\nwith the advantages of cloud-based infrastructure described\nin the previous section (\u201cBisQue for Cloud Based Data Materials and Data Collection\nStorage and Analysis\u201d). The DREAM.3D processing capa-\nbilities were also extended to allow for parallelized pro- The 3D datasets were collected via TriBeam tomography\ncessing across wide parameter spaces. Anytime a process- from two additively manufactured alloys in their as-\ning pipeline is generated in DREAM.3D through BisQue, built state, meaning no additional heat treatments were\nparameter input variables can be represented as a range performed on the samples after additive processing.\nof values (with sub-increments), rather than as a distinct TriBeam tomography [33] utilizes a femtosecond laser\nvalue. An example of DREAM.3D pipeline execution is for micromachining within a dual-beam scanning electron\nshown in Fig. 2 alongside the analogous representation microscope (SEM) equipped with a focused ion beam\nin DREAM.3D. The user selects an input source and (FIB). The ultrashort pulse length minimizes damage to\npipeline source, parameterizable values are then automati- the machined surface while enabling high material removal\ncally displayed in the parameter section. Using straightfor- rates, roughly four to five orders of magnitude faster than\nward list comprehension, many independent instantiations FIB milling alone, resulting in an accessible volume on\nof DREAM.3D can be run at once across multiple nodes on the order of a cubic millimeter [30, 34]. The FIB can\na compute cluster, dramatically decreasing setup and analy- also be used at a glancing angle to minimize surface\nsis time for detailed parameter studies. The resulting output roughness due to laser-induced periodic surface structures\nfiles can then be examined as individual pipelines as BisQue (LIPSS); however, this cleaning step is material dependent\nhas a fully integrated table viewer that allows for both table and is not always required to obtain high-quality 3D EBSD\nviewing and inspection of the tree structure of HDF files. data.40 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\n\n\n\nFig. 2 DREAM.3D pipeline for reconstruction of an additively manu- parameterization for variable inputs in a DREAM.3D pipeline, which\nfactured 304L Stainless Steel 3D dataset as viewed in (a) DREAM.3D can be executed in parallel on high-performance compute clusters and\non a workstation computer and (b) the DREAM.3D BisQue mod- then analyzed to determine the ideal reconstruction parameters\nule in a web browser. The DREAM.3D BisQue module allows for\n\n\n\nLENS-Processed 304L Stainless Steel EBSD scans were collected on each slice with an in-plane\n resolution of 3 \u03bcm, yielding square voxels. Roughly 200 GB\nLaser engineered net shaping (LENS) is a directed energy of data was collected from a total of 80 slices over the course\ndeposition (DED) additive manufacturing technique derived of two days, resulting in a volume of 816 \u00d7 834 \u00d7 240 \u03bcm.\nfrom laser cladding processes, in which metal powder Raw EBSD patterns, comprising nearly 90% of the collected\nis blown into a molten pool created using a laser. The data, were saved for post-processing. Individual slices were\nlow dilution area and high deposition rates achievable in reconstructed into a coherent volume using DREAM.3D\nLENS, up to an order of magnitude higher than powder- in BisQue. A minimum confidence index (CI) of 0.1 was\nbed additive processes, allow for the manufacture of large chosen as the threshold for data quality, as the indexing\nscale parts [35]. A sample of 304L stainless steel fabricated of EBSD patterns has been shown to be 95% accurate at\nby the LENS process was characterized in the TriBeam a CI of 0.1 [36]. This threshold is then used to create a\nsystem using a slice thickness of 3 \u03bcm. The femtosecond \u201cmask\u201d that flags individual voxels as containing good or\nlaser ablated surface was cleaned using glancing-angle bad (noisy) data, so the mask represents a subset of the\nFIB milling to increase the quality of EBSD patterns. total volume characterized during data collection. The maskIntegr Mater Manuf Innov (2019) 8:37\u201351 41\n\n\nfor the sample of 304L can be seen in Fig. 3. Grains were were saved for post-processing. Glancing-angle FIB was not\nsegmented using a misorientation tolerance as described required to resolve EBSD patterns, which were collected\nin \u201cMisorientation Tolerance Parameterization for Feature with an in-plane resolution of 1.5 \u03bcm to create square\nIdentification,\u201d and grains containing less than 64 voxels voxels. As with the 304L sample, a minimum CI of 0.1\nwere considered too small to be resolved and were removed. was chosen for the initial mask. Grains were segmented\nThis corresponds to 4 \u00d7 4 \u00d7 4 points across a feature and an using a misorientation tolerance of 2\u25e6using the same\nequivalent diameter of 14.9 \u03bcm, roughly five times smaller segmentation algorithm as explained in \u201cMisorientation\nthan the average grain size determined from 2D analysis Tolerance Parameterization for Feature Identification.\u201d A\ndone prior to 3D characterization. Gaps remaining from minimum size requirement of 125 voxels was applied to\nthese removed features were filled via iterative dilation of delete grai\n\n## 6. polonsky2019threedimensionalanalysisand pages 1-2 (Context ID: pqac-00000011)\n\nIntegrating Materials and Manufacturing Innovation (2019) 8:37\u201351\nhttps://doi.org/10.1007/s40192-019-00126-7\n\n TECHNICAL ARTICLE\n\n\nThree-dimensional Analysis and Reconstruction of Additively\nManufactured Materials in the Cloud-Based BisQue Infrastructure\n\nAndrew T. Polonsky1 \u00b7 Christian A. Lang2 \u00b7 Kristian G. Kvilekval2 \u00b7 Marat I. Latypov1,3 \u00b7 McLean P. Echlin1 \u00b7\nB. S. Manjunath2 \u00b7 Tresa M. Pollock1\n\n\nReceived: 30 December 2018 / Accepted: 18 February 2019 / Published online: 22 March 2019\n\u00a9 The Minerals, Metals & Materials Society 2019\n\nAbstract\nA microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a\ncloud-based platform, BisQue. Parallelization of DREAM.3D module executions and the ability to parameterize pipeline\nvariables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-\ncollected 3D EBSD datasets of additively manufactured materials with complex anisotropic microstructures. Furthermore,\na comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using\nBisQue\u2019s parallelized DREAM.3D module executions. The direction of cloud-based data infrastructure and the prospects\nfor impact in material science are also discussed.\n\nKeywords Tomography \u00b7 3D materials science \u00b7 Microstructure analytics \u00b7 Cloud infrastructure \u00b7 Additive manufacturing \u00b7\n3D metal printing\n\n\n Introduction\n Andrew T. Polonsky\n polonsky@ucsb.edu\n Recent external pressures mandated by government grants\n Christian A. Lang requiring data be accessible to the public, and internal\n clang@ucsb.edu pressures within academic communities for data sharing\n and reproducibility have led to the emergence of data-\n Kristian G. Kvilekval\n driven platforms within the scientific community [1\u20134].\n kgk@ucsb.edu\n Requirements for these platforms include providing data\n Marat I. Latypov provenance, data sharing, and hosting services for data that\n latmarat@ucsb.edu may otherwise disappear when a lab closes or hardware\n fails [5]. Recently, data repositories have shifted from\n McLean P. Echlin\n mechlin@ucsb.edu simply siloing data to include data analytics and predictive\n capabilities. Furthermore, some of these repositories are\n B. S. Manjunath directly linked with high-performance computing in order\n manj@ucsb.edu\n to populate an ever growing database, such as with\n Tresa M. Pollock thermodynamic databases such as the Materials Project [6]\n tresap@ucsb.edu and the Open Quantum Materials Database (OQMD) [7].\n More sophisticated predictive/interactive databases, such as\n1 the aforementioned thermodynamic ones, have been slower Materials Department, University of California Santa Barbara,\n Santa Barbara, CA 93106, USA to develop with the 3D Materials Science community. This\n2 is likely due to the less standardized and more diverse data Electrical and Computer Engineering Department,\n University of California Santa Barbara, Santa Barbara, modalities, the complexity and diversity of microstructures,\n CA 93106, USA and the relative infancy of the community itself.\n3 Mechanical Engineering Department, University of California In almost all regards, using 3D data to answer scientific\n Santa Barbara, Santa Barbara, CA 93106, USA questions in Materials Science is challenging due to the38 Integr Mater Manuf Innov (2019) 8:37\u201351\n\n\nneed for dataset acquisition, reconstruction, and analysis. In 3D characterization techniques, such as X-ray-based\nReconstruction of 3D datasets is a complex process with diffraction contrast tomography (DCT), the problems of\nmany degrees of freedom. The algorithms employed, and data reconstruction and analysis are also convoluted. Coher-\ntolerances selected by the user to align slices, segment ent volumes must be reconstructed from individual 2D\nfeatures, and remove data artifacts (i.e., remove noise), images via back projection techniques [24]. The diffraction\nultimately determine the quantitative data extracted from spots from the 2D projections are generally assumed to orig-\nthe volume. Significant progress on the reconstruction and inate from a single grain, limiting the characterization of\nanalysis of electron backscatter diffraction (EBSD) data has materials with large internal strain gradients [25]. Moreover,\nbeen made by a software package DREAM.3D [8], which stereological conventions established in 2D do not neces-\ncreates a unified data structure for 3D data treatment and sarily capture the real 3D microstructures, particularly at\naggregates tools and algorithms collectively developed by extrema of property distributions [26, 27]. Given the wealth\nthe community. In DREAM.3D, this process takes the form of additional data provided by 3D characterization, new\nof a pipeline, which typically consists of well over a dozen tools aimed at exploring the parameter space offered to 3D\nsteps, and in which each step has its own set of input reconstruction are required. In the following manuscript, we\nvariables and outputs. DREAM.3D typically runs in a serial describe the use of a cloud-based platform, BisQue [28, 29],\nprocess on a workstation, with predefined input parameters. enhanced with the DREAM.3D microstructural software\nDue to the long computation times required to process analysis package for the 3D reconstruction and analysis\n3D data, which often scales proportionately as dataset size of additively manufactured microstructures characterized\nand complexity increases, it is impractical to explore the using TriBeam tomography [30].\nfull parameter space by manually running reconstructions.\nTypically, only several variables are adjusted by a user until\na qualitatively acceptable reconstruction is attained. This Methods\nprocess inherently imparts bias and typically provides little\nopportunity to justify specific reconstruction variables. The following sections describe the software infrastructure\n The bias generated during reconstruction of 3D datasets and materials investigated.\nis subsequently carried through to data analysis, which,\nfor microstructural data, is still an ill-defined process. BisQue for Cloud-Based Data Storage and Analysis\nStandardization of analysis involving new characterization\ntechniques is generally a slow process. Automated orien- All data processing was performed in the cloud with BisQue\ntation imaging analysis via EBSD first emerged in the [28, 29] using a web browser interface to interact with the\nearly 90s, but standards did not emerge for nearly two 3D datasets. BisQue has backend infrastructure support for\ndecades [9, 10]. These new standards covered basic quan- web services such as Amazon S3, IRODS, and federated\ntitative analysis, including methods for grain size measure- local file systems as well as client facing tools such as\nment, as well as guidelines governing data quality, quan- Java, Python, and Matlab. A DREAM.3D [8] plugin was\ntity, and resolution sufficient to make such measurements developed for BisQue in order to incorporate the materials\n[11]. Even in 2D EBSD analysis, such standards are rare science based 3D dataset analysis tools. The database\nand do not necessarily address additional subtleties arising architecture of BisQue is such that every processing\nfrom more complex, albeit commonly studied, microstruc- execution, dataset (input and output), and workflow\ntures, including twinned materials, nanocrystalline mate- processing pipeline receives a unique web address URL.\nrials, and deformed materials [12\u201317]. Standardization of These URLs are useful for at least two reasons: tracking\ncharacterization meth\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011]\n\nStatus: Paper Count=22 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=12 | Current Cost=$0.1324 | Disease-Target Evidence=0", + "name": "read_text", + "tool_call_id": "call_1oiOo0bOuigk4ZUPXbdj7xZ2", + "info": { + "start_ts": 805.142754992, + "end_ts": 819.6230742 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_eVCSSI1d7fc6pWTEoIMUSTF0", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 1, \"updated_plan_objective_json\": \"{\\\"id\\\":1,\\\"objective\\\":\\\"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\\\",\\\"rationale\\\":\\\"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\\\",\\\"evaluation\\\":\\\"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_eVCSSI1d7fc6pWTEoIMUSTF0", + "info": { + "start_ts": 845.159035053, + "end_ts": 845.160391077 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_UpEp4rBNUwkVd82OjFFGqJTn", + "type": "function", + "function": { + "arguments": "{\"question\": \"What peer-reviewed evidence establishes MTEX as a benchmark for EBSD grain detection/reconstruction and crystallographic texture analysis, including ODF, pole figures, IPF, and advanced twin reconstruction? What cautions apply to misorientation-threshold grain segmentation?\"}", + "name": "gather_evidence" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Most relevant evidence (with ids):\n\n## pqac-00000012\n\nFrom text 'savage2021anautomatedprocedure pages 1-2'.\n\nDaniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\nThis peer-reviewed paper (Materials Characterization, 2021) demonstrates MTEX's capability for advanced EBSD analysis, specifically automated reconstruction of deformation twin hierarchies in heavily twinned microstructures. The work addresses critical challenges in twin identification including missing twin relationships, grouping of deformed grain fragments, and classification of multi-family twin relationships. The authors developed algorithms combining MTEX with MATLAB graph toolboxes to significantly improve automation compared to existing methodologies. The procedure successfully identifies third-generation twinning in \u03b1-titanium and can extract twin statistics at larger deformation levels than previously possible. However, the excerpt does not provide information about MTEX's performance regarding ODF, pole figures, or IPF analysis, nor does it discuss cautions regarding misorientation-threshold grain segmentation methods.\n\n\n---\n\n## pqac-00000013\n\nFrom text 'savage2021anautomatedprocedure pages 3-4'.\n\nDaniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\nThis peer-reviewed article from Materials Characterization demonstrates MTEX's application for grain reconstruction and texture analysis in heavily twinned \u03b1-Ti microstructures. The authors use MTEX's texture analysis capabilities to handle complex microstructures with large twin volume fractions, multiple twin variants, and higher-order twinning. The paper presents inverse pole figure (IPF) EBSD maps and discusses texture analysis with strong c-axis texture observations. Regarding misorientation-threshold grain segmentation cautions, the authors note that due to strong texture, neighboring grains often have low-angle boundaries, and twins commonly chain across grain boundaries with only small misorientation differences. They address this by biasing reconstruction towards over-segmentation rather than under-segmentation and using different misorientation tolerances (1-2\u00b0) depending on loading direction. The authors acknowledge that under-segmentation and low-angle boundary treatment presents challenges in grain reconstruction.\n\n\nArtifact IDs that may be injected into the answer: []\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013]\n\nStatus: Paper Count=22 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=14 | Current Cost=$0.1716 | Disease-Target Evidence=0", + "name": "gather_evidence", + "tool_call_id": "call_UpEp4rBNUwkVd82OjFFGqJTn", + "info": { + "start_ts": 868.044797633, + "end_ts": 874.62140145 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_P8jQvXEYp1OGYeb95uE317Pf", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 2, \"updated_plan_objective_json\": \"{\\\"id\\\":2,\\\"objective\\\":\\\"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\\\",\\\"rationale\\\":\\\"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\\\",\\\"evaluation\\\":\\\"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_P8jQvXEYp1OGYeb95uE317Pf", + "info": { + "start_ts": 898.53099696, + "end_ts": 898.532407154 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_QJeM4kTCkLbZRr827PUgMWr1", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 3, \"updated_plan_objective_json\": \"{\\\"id\\\":3,\\\"objective\\\":\\\"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\\\",\\\"rationale\\\":\\\"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\\\",\\\"evaluation\\\":\\\"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_QJeM4kTCkLbZRr827PUgMWr1", + "info": { + "start_ts": 923.49329889, + "end_ts": 923.494253073 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_ibuikCX928XmGVA9Di78lliY", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 4, \"updated_plan_objective_json\": \"{\\\"id\\\":4,\\\"objective\\\":\\\"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\\\",\\\"rationale\\\":\\\"The answer should support a concrete implementation decision, not merely list software.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\\\",\\\"evaluation\\\":\\\"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"completed\",\n \"result\": \"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\",\n \"evaluation\": \"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_ibuikCX928XmGVA9Di78lliY", + "info": { + "start_ts": 957.818000802, + "end_ts": 957.820227968 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_FRljLsn8HFS1KSTBdRAWzMFk", + "type": "function", + "function": { + "arguments": "{\"artifact_type\": \"table\", \"description\": \"Create a comparison matrix for MTEX, kikuchipy, orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, and OpenXY/research scripts. Columns: package/ecosystem; Python-native status; grain reconstruction/segmentation; texture ODF/PF/IPF; KAM/GND; maturity/reference status; recommended role. Use concise capability ratings such as strong, partial/custom, absent/not core, and clearly avoid claiming kikuchipy or pyxem are complete MTEX replacements.\"}", + "name": "create_artifact" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "### Artifact (ID artifact-00)\n\n| Package / ecosystem | Python-native status | Grain reconstruction / segmentation | Texture ODF / PF / IPF | KAM / GND | Maturity / reference status | Recommended role |\n|---|---|---|---|---|---|---|\n| MTEX | No; MATLAB-based | **Strong** for symmetry-aware 2D grain detection/reconstruction; advanced specialized workflows documented | **Strong**; benchmark for ODF, pole figures, IPF, texture components (pqac-00000000, pqac-00000012, pqac-00000013) | **Strong** in the MTEX ecosystem; widely treated as benchmark | Very mature, heavily cited benchmark; peer-reviewed advanced EBSD workflows documented (pqac-00000012, pqac-00000013) | Functional benchmark and validation target, but not MATLAB-free |\n| kikuchipy | Yes | **Absent / not core** for downstream grain reconstruction; primarily EBSD pattern processing/indexing support | **Partial / custom** via interoperability, not a full texture-analysis suite | **Absent / not core** | Open-source Python EBSD toolbox with literature record, but not a complete post-processing replacement in the retrieved evidence | Upstream pattern handling, preprocessing, indexing, data I/O; pair with orix and/or DREAM3D-NX |\n| orix | Yes | **Partial / custom**; orientation objects and misorientation logic support custom grain segmentation, but not MTEX-level turnkey reconstruction | **Strong / near-benchmark for orientation visualization**; IPF/PF native, ODF possible but typically less turnkey than MTEX | **Partial / custom**; KAM-like metrics and curvature workflows need user implementation/validation | Peer-reviewed library record available; positioned as crystallographic orientation library rather than full EBSD suite | Core Python orientation engine for custom EBSD post-processing and visualization |\n| DefDAP | Yes | **Partial**; useful 2D deformation/EBSD map grain analysis, but narrower than MTEX | **Partial**; some crystallographic plotting/analysis, not a full MTEX texture environment | **Partial**; deformation-oriented metrics are a strength, GND generally custom/limited | Open-source Python project with software-release citation; maturity moderate, community smaller than MTEX/DREAM3D | Good lightweight Python choice for 2D grain/deformation analysis after indexing |\n| DREAM3D-NX / simplnx | Yes (Python API around DREAM3D-NX) | **Strong**; mature pipeline-based segmentation/reconstruction, especially for 2D/3D EBSD and batch automation (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000011) | **Partial**; some crystallographic statistics, but not as texture-centric or interactive as MTEX (pqac-00000004) | **Partial / custom**; not typically the first choice for KAM/GND-centric analysis | Mature EBSD/microstructure pipeline ecosystem with peer-reviewed reconstruction studies (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Best Python-accessible option for robust automated grain segmentation/statistics in a MATLAB-free pipeline |\n| pyxem | Yes | **Absent / not core for SEM-EBSD**; focused on diffraction microscopy in TEM/4D-STEM, not a direct MTEX replacement | **Partial** for orientation mapping/visualization in diffraction-microscopy contexts | **Absent / not core** | Mature open-source microscopy library, but its center of gravity is 4D-STEM/SPED rather than conventional SEM-EBSD post-processing | Use only if your workflow spans diffraction microscopy; not recommended as primary EBSD grain-texture backend |\n| PyMicro | Yes | **Partial / custom**; useful microstructure data structures and crystallography, but less turnkey for EBSD grain reconstruction | **Partial / custom** | **Partial / custom** | Useful research software ecosystem, but less established as a mainstream EBSD post-processing stack in the retrieved evidence | Supporting library for custom workflows, data models, simulation coupling |\n| OpenXY / research scripts / pyEBSD-style code | Yes, but heterogeneous | **Partial / custom** to **strong in isolated tasks**; highly variable and often method-specific | **Partial / custom** | **Partial / custom**, including possible GND research implementations | Lowest standardization; often lacks broad peer-reviewed software validation or stable APIs | Fill specific gaps such as KAM/GND or custom segmentation, but validate carefully against MTEX/DREAM3D outputs |\n| Practical no-MATLAB stack for AM AlSi10Mg | Yes | **Strong overall** if composed as: kikuchipy/PyEBSDIndex + DREAM3D-NX for segmentation + orix/DefDAP for orientation/statistics | **Partial-to-strong overall**; strongest with orix plus custom ODF tooling | **Partial / custom overall**; KAM feasible, GND should be a separately validated module | Most realistic MATLAB-free path based on current open-source ecosystem and evidence balance (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Recommended integrated pipeline rather than searching for a single Python package that fully replaces MTEX |\n\n\n*Table: This table compares major open-source Python-native or Python-accessible EBSD post-processing options against MTEX for grain reconstruction, texture analysis, and misorientation-derived metrics. It is designed to help choose a MATLAB-free downstream pipeline for automated EBSD workflows.*\n\nArtifact IDs that may be injected into the answer: [artifact-00]\n\nContext IDs that may be cited: [pqac-00000000, pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000006, pqac-00000007, pqac-00000008, pqac-00000009, pqac-00000010, pqac-00000011, pqac-00000012, pqac-00000013]\n\nStatus: Paper Count=22 | Relevant Papers=6 | Clinical Trial Count=0 | Relevant Clinical Trials=0 | Current Evidence=14 | Current Cost=$0.1716 | Disease-Target Evidence=0", + "name": "create_artifact", + "tool_call_id": "call_FRljLsn8HFS1KSTBdRAWzMFk", + "info": { + "start_ts": 980.285662375, + "end_ts": 993.711719606 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_7bcoFGw2YuGzai5ip8f8B7P4", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 5, \"updated_plan_objective_json\": \"{\\\"id\\\":5,\\\"objective\\\":\\\"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\\\",\\\"rationale\\\":\\\"A structured matrix will make the multi-package comparison actionable.\\\",\\\"status\\\":\\\"completed\\\",\\\"result\\\":\\\"Created artifact-00, a package-by-package matrix covering Python status, grains, texture, KAM/GND, maturity, and recommended role.\\\",\\\"evaluation\\\":\\\"The matrix directly supports package selection and clearly distinguishes full, partial, and non-core capabilities.\\\"}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"completed\",\n \"result\": \"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\",\n \"evaluation\": \"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a package-by-package matrix covering Python status, grains, texture, KAM/GND, maturity, and recommended role.\",\n \"evaluation\": \"The matrix directly supports package selection and clearly distinguishes full, partial, and non-core capabilities.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"pending\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_7bcoFGw2YuGzai5ip8f8B7P4", + "info": { + "start_ts": 1015.613473393, + "end_ts": 1015.614209595 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_0Ow8AsXKD75JrGG2nzDUQk6Q", + "type": "function", + "function": { + "arguments": "{\"plan_objective_id\": 6, \"updated_plan_objective_json\": \"{\\\"id\\\":6,\\\"objective\\\":\\\"Call the answer tool with a thorough, evidence-cited technical response.\\\",\\\"rationale\\\":\\\"This is required to submit the final answer after all research objectives are complete.\\\",\\\"status\\\":\\\"in-progress\\\",\\\"result\\\":null,\\\"evaluation\\\":null}\"}", + "name": "update_plan" + } + } + ], + "info": null + }, + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"completed\",\n \"result\": \"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\",\n \"evaluation\": \"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a package-by-package matrix covering Python status, grains, texture, KAM/GND, maturity, and recommended role.\",\n \"evaluation\": \"The matrix directly supports package selection and clearly distinguishes full, partial, and non-core capabilities.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_0Ow8AsXKD75JrGG2nzDUQk6Q", + "info": { + "start_ts": 1037.185396381, + "end_ts": 1037.186815636 + } + }, + { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_8LdlgnIRhAHn08c2v67pYdEC", + "type": "function", + "function": { + "arguments": "{\"answer\": \"# Open-source Python options for EBSD post-processing\\n\\n## Executive conclusion\\n\\nThere is currently **no single Python-native package that reproduces the full integrated scope and maturity of MTEX**. The most credible MATLAB-free solution is a composed stack:\\n\\n1. **kikuchipy + PyEBSDIndex** for pattern processing and indexing;\\n2. **DREAM3D-NX through `simplnx`** for reproducible grain segmentation, cleanup, grain statistics, and especially 3D-capable workflows;\\n3. **orix** as the crystallographic orientation, symmetry, misorientation, IPF, and pole-figure layer;\\n4. optionally **DefDAP** for convenient 2D grain/deformation-map analysis;\\n5. a separately tested module for **KAM**, and especially for **GND-density inversion**.\\n\\nFor the proposed automated AlSi10Mg pipeline, DREAM3D-NX/`simplnx` is the strongest grain-characterization backend, while orix is the best Python-native crystallographic foundation. kikuchipy itself is not a downstream replacement for MTEX, and pyxem is primarily a diffraction-microscopy/4D-STEM package rather than a conventional SEM-EBSD grain-analysis suite.\\n\\n\\n\\n## The MTEX benchmark\\n\\nMTEX combines, in one environment, symmetry-aware EBSD import and manipulation, grain reconstruction, grain and boundary objects, texture estimation and plotting, misorientation statistics, KAM/GROD-type measures, and curvature/dislocation-density workflows. Its texture functionality includes ODF estimation, pole and inverse-pole figures, texture components, fibres, and quantitative texture measures. Published applications use MTEX for IPF maps, ODF-derived pole figures, grain-size and grain-shape statistics, and grain-boundary analysis. (pqac-00000000)\\n\\nMTEX also supports sophisticated extensions beyond simple threshold segmentation. A peer-reviewed workflow built on MTEX and MATLAB graph tools reconstructed deformation-twin hierarchies, grouped deformed fragments into parent families, treated missing relationships, and identified third-generation twins in heavily twinned \\u03b1-Ti. (pqac-00000009, pqac-00000012) This illustrates the principal gap in Python: individual Python libraries provide many underlying operations, but MTEX supplies a more coherent grain/boundary/texture abstraction and a larger collection of specialized reconstruction workflows.\\n\\nMTEX is not immune to methodological choices. Strong textures and low-angle boundaries can cause under-segmentation or chains of crystallographically similar regions; one MTEX study deliberately biased reconstruction toward over-segmentation and used 1\\u20132\\u00b0 tolerances for its particular material. (pqac-00000013) More generally, grain counts and geometry depend strongly on the selected threshold, connectivity, cleanup, and minimum-grain-size rules. A conventional 5\\u00b0 threshold is common but is not universally valid. (pqac-00000008)\\n\\n## Package assessment\\n\\n### 1. kikuchipy\\n\\n**Role.** kikuchipy is the natural upstream package in the proposed workflow. It is designed around EBSD patterns: vendor-file access, lazy handling of large pattern arrays, detector and projection geometry, background correction, pattern processing, simulation-related operations, and indexing interoperability. Its data model uses the HyperSpy ecosystem and orix crystallographic objects.\\n\\n**Grains and texture.** It should not be selected as the principal grain-reconstruction engine. Once indexing has produced orientation, phase, quality, and coordinate arrays, downstream segmentation and grain statistics generally require another package or custom code. Likewise, kikuchipy is not a comprehensive ODF/texture-analysis system and does not offer an MTEX-equivalent grain/boundary object model.\\n\\n**KAM/GND.** Neither is a core turnkey kikuchipy capability. KAM can be calculated from its output arrays with orix/NumPy-style operations, but masking, symmetry reduction, neighborhood definition, and exclusion of high-angle boundaries remain the pipeline developer\\u2019s responsibility. GND inversion is substantially further outside its main scope.\\n\\n**Maturity.** kikuchipy is a well-developed, domain-specific open-source project with a software paper titled *kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns*. Its maturity is strongest for pattern-space operations, not for post-indexing microstructure analysis.\\n\\n### 2. orix\\n\\n**Role.** orix is the most important Python-native crystallographic foundation. It supplies rotations and orientations, crystal and specimen symmetry, fundamental-zone reduction, symmetry-aware misorientation, orientation colours, inverse-pole-figure colour keys, pole-figure-style projections, and orientation clustering. Its peer-reviewed methodological reference is Johnstone et al., *Density-based clustering of crystal (mis)orientations and the orix Python library*, *Journal of Applied Crystallography* 53 (2020), 1293\\u20131298, DOI 10.1107/S1600576720011103.\\n\\n**Grain reconstruction.** orix contains the primitives needed to compare neighboring orientations and cluster orientation data, but it is not yet an MTEX-like turnkey EBSD topology package. A production grain segmenter still needs scan-grid connectivity, phase masks, confidence masks, flood-fill or graph labeling, small-region handling, and grain-property aggregation. Those steps can be implemented with NumPy, SciPy, scikit-image, xarray, and networkx, but then the laboratory owns validation and maintenance.\\n\\n**Texture.** orix is strong for IPF colouring, stereographic projections, symmetry handling, and discrete orientation distributions. It is suitable for PF/IPF visualization and custom orientation-density estimation. Nevertheless, MTEX remains substantially richer and more established for quantitative ODF estimation, kernel-selection conventions, harmonic representations, fibres/components, uncertainty-aware texture workflows, and publication-ready texture analysis. An ODF reconstructed from discrete EBSD orientations is also parameter-dependent; published work notes sensitivity to ODF computation settings in MTEX itself, so Python/MTEX agreement requires matching kernels, bandwidths, symmetry, weights, and normalization. A modern EBSD texture study used pole figures to characterize weak texture and illustrates the need to retain a statistically representative orientation population. (pqac-00000006)\\n\\n**KAM/GND.** orix is a good basis for KAM because it provides symmetry-aware misorientation operations. It does not, by itself, turn KAM into a one-call validated EBSD workflow. GND requires spatial orientation gradients, reference-frame consistency, Burgers vectors/slip-system assumptions, Nye-tensor construction, and an underdetermined inversion; this remains custom research code rather than an MTEX-equivalent orix feature.\\n\\n### 3. DefDAP\\n\\n**Role.** DefDAP\\u2014Deformation Data Analysis in Python\\u2014is a Python package aimed at combining EBSD with deformation measurements such as high-resolution digital image correlation. Its citable software release is Atkinson et al., *DefDAP: Deformation Data Analysis in Python* (2020), DOI 10.5281/zenodo.3688097.\\n\\n**Grains.** DefDAP is useful for 2D EBSD maps, grain finding and grain-level analysis, crystallographic slip/twinning interpretation, boundary overlays, and linking deformation fields to grains. For a conventional single-phase 2D Al EBSD map it may be the quickest lightweight route from an orientation map to grains and grain-wise quantities.\\n\\n**Texture and misorientation metrics.** Its emphasis is local deformation and grain-scale mechanics, not comprehensive quantitative texture. It is less complete than MTEX for ODFs, texture components, and generalized boundary/reconstruction operations. KAM-like local misorientation calculations can be added or may be available through map-level operations depending on release, but GND-density inversion should not be assumed to be a complete, standardized capability without checking the exact version and validating the implementation.\\n\\n**Maturity.** DefDAP has been used in peer-reviewed deformation studies, but its community, documentation breadth, and EBSD feature coverage are smaller than MTEX or DREAM.3D. It is best regarded as a focused 2D deformation-analysis package rather than a universal EBSD backend.\\n\\n### 4. DREAM.3D and DREAM3D-NX/`simplnx`\\n\\n**Role and architecture.** Legacy DREAM.3D is an established C++/Qt microstructure-analysis platform. DREAM3D-NX is its current-generation successor; `simplnx` provides Python bindings and pipeline construction. Thus, it is **Python-accessible and scriptable**, although not pure Python internally. This distinction is normally beneficial for large EBSD maps because segmentation and feature-statistics filters execute in compiled code.\\n\\n**Grain reconstruction.** This is the strongest open-source alternative for automated grain segmentation. DREAM.3D uses pipeline filters for data import, masking, cleanup, symmetry-aware misorientation segmentation, feature identification, neighbor lists, grain morphology, crystallographic statistics, boundary meshes, and, where applicable, 3D reconstruction. Published work describes DREAM.3D as a unified data structure aggregating community-developed algorithms, with reconstruction pipelines commonly containing more than a dozen parameterized steps. (pqac-00000011)\\n\\nIts `Segment Features (Misorientation)` workflow has been used to compare 2D and 3D segmentation across several misorientation thresholds. (pqac-00000007) Published AM/TriBeam workflows reconstructed serial EBSD slices, segmented grains using explicit misorientation tolerances, removed features below specified voxel counts, filled gaps, and automated parameter sweeps. (pqac-00000003, pqac-00000010) Cloud integration further demonstrated parallel parameter studies, provenance tracking, and grain-size comparisons between 2D and 3D AM datasets. (pqac-00000001, pqac-00000002)\\n\\n**Texture.** DREAM.3D calculates crystallographic feature statistics and supports orientation-distribution data used in synthetic microstructure generation, but it is not as strong as MTEX for interactive quantitative texture analysis. PF/IPF maps and basic orientation statistics are feasible; sophisticated ODF reconstruction and texture-component analysis are better handled in orix plus custom numerical code\\u2014or validated against MTEX. Published workflows frequently use DREAM.3D for microstructure generation/reconstruction and MTEX for detailed ODF/PF analysis, reflecting this division of strengths. (pqac-00000000)\\n\\n**KAM/GND.** DREAM3D-NX should not be selected primarily for KAM or GND. Depending on installed plugins and version, local orientation filters may cover some neighborhood measures, but these capabilities are less central and less transparent than its feature segmentation/statistics pipeline. A NumPy/orix KAM stage is easier to audit. GND remains a specialist calculation.\\n\\n**Cautions.** Segmentation is not parameter-free. The same threshold can produce different 2D and 3D connectivity, especially for nonconvex grains or paths connected outside the selected plane. (pqac-00000005, pqac-00000007, pqac-00000008) Minimum feature size and cleanup can also shift the measured grain-size distribution. Consequently, pipeline JSON, threshold, neighborhood convention, phase mask, and deleted/fill operations should be retained as provenance.\\n\\n### 5. pyxem\\n\\npyxem is an open-source Python ecosystem for multidimensional diffraction microscopy, built around HyperSpy and commonly used for scanning/precession electron diffraction and 4D-STEM. It is valuable for virtual diffraction imaging, diffraction-vector analysis, orientation mapping, and related TEM workflows.\\n\\nIt is **not a direct downstream package for conventional SEM-EBSD orientation maps**. Grain segmentation from vendor-style EBSD grids, quantitative EBSD texture analysis, KAM, and GND are not its central abstractions. Its shared ecosystem with kikuchipy/orix can be useful if the project later combines EBSD with scanning electron diffraction, but pyxem should not be the primary AlSi10Mg grain-analysis backend.\\n\\n### 6. PyMicro\\n\\nPyMicro provides Python classes and algorithms for crystallography, microstructure representation, grains, orientations, meshes, and links to simulation and tomography. It can import or represent EBSD-like microstructures and is useful for custom data integration, synthetic microstructures, and computational-mechanics coupling.\\n\\nFor this task, its limitations are that conventional EBSD segmentation, comprehensive ODF analysis, KAM, and GND are not offered with the same turnkey breadth or community validation as MTEX. It is better viewed as an extensible microstructure framework than as a drop-in EBSD analysis application.\\n\\n### 7. OpenXY and other research code\\n\\nOpenXY is associated mainly with high-angular-resolution EBSD cross-correlation and strain/rotation analysis. It addresses a different problem from ordinary Hough-indexed orientation-map characterization and is not a complete grain/texture package. Some Python ports and research repositories calculate lattice curvature or GND density, but their conventions, slip-system bases, regularization, and maintenance vary.\\n\\nOther small packages carrying names such as `pyebsd`, custom scikit-image segmenters, or laboratory scripts can calculate IPF colours, KAM, or threshold-connected grains. They can be useful building blocks, but package names are sometimes ambiguous, releases may be unmaintained, and peer-reviewed software references are often absent. They should not be treated as equivalent to MTEX merely because they produce an IPF or KAM image.\\n\\n## Recommended pipeline for AM AlSi10Mg\\n\\n### Data contract after indexing\\n\\nExport or retain the following arrays and metadata:\\n\\n- quaternion or rotation matrix per scan point\\u2014preferably not only Euler angles;\\n- phase ID, indexed/unindexed mask, confidence or match score, and pattern quality;\\n- physical `x`, `y` coordinates and step size;\\n- scan shape and rectangular/hexagonal topology;\\n- crystal symmetry and lattice parameters for fcc Al;\\n- detector/sample/reference-frame conventions;\\n- indexing dictionary or phase-library version.\\n\\nAlSi10Mg is commonly dominated in EBSD by fcc \\u03b1-Al, while the fine Si network may be incompletely indexed at ordinary step sizes. Do not let unindexed Si-rich pixels automatically split Al grains: establish an explicit phase/mask and cleanup policy and report it.\\n\\n### Grain stage\\n\\nUse **DREAM3D-NX/`simplnx`** for the production pipeline:\\n\\n1. construct the image geometry and load quaternion, phase, mask, and quality arrays;\\n2. remove only demonstrably unreliable points;\\n3. segment same-phase connected pixels with a documented disorientation threshold;\\n4. calculate feature centroids, equivalent diameters, areas, aspect ratios, mean orientations, neighbors, and boundary misorientations;\\n5. retain both the raw feature labels and any post-cleanup labels;\\n6. sweep plausible thresholds rather than fixing one unexamined value.\\n\\nFor AM AlSi10Mg, evaluate at least a low-angle-sensitive threshold and a conventional grain threshold\\u2014for example, a controlled sweep around 2\\u20135\\u00b0\\u2014because cellular/subgrain orientation gradients may be physically meaningful. This is not a recommendation to choose a universal value; published work demonstrates that changing tolerance can radically alter segmentation and that 2D and 3D results are not automatically equivalent. (pqac-00000003, pqac-00000005, pqac-00000007)\\n\\nIf only modest 2D maps are involved and direct Python object manipulation is more important than DREAM3D-NX pipelines, **DefDAP** is a reasonable alternative. A fully custom orix/scikit-image segmenter is defensible only if regression-tested against DREAM3D-NX or MTEX.\\n\\n### Texture stage\\n\\nUse **orix** for orientation conversion, symmetry reduction, IPF maps, pole figures, and discrete orientation statistics. For ODF estimation:\\n\\n- weight points or grains deliberately; pixel-weighted and one-orientation-per-grain ODFs answer different questions;\\n- state the kernel and half-width/bandwidth;\\n- enforce cubic crystal symmetry and the intended specimen symmetry;\\n- normalize PF/ODF intensities consistently, usually in multiples of random distribution;\\n- validate against a known texture or a small MTEX reference dataset.\\n\\nThis distinction matters in AM material: pixel weighting emphasizes large columnar grains, whereas grain weighting describes the population of grains. A recent study highlights the importance of preserving the full orientation distribution when reducing EBSD orientation datasets and uses PF maps to assess texture strength. (pqac-00000006)\\n\\n### KAM\\n\\nKAM is straightforward to implement but easy to define inconsistently. For each indexed point, compute the mean symmetry-reduced disorientation to valid neighbors, while specifying:\\n\\n- first-, second-, or higher-order neighborhood;\\n- whether diagonals are included;\\n- maximum included disorientation, often chosen to exclude grain boundaries;\\n- treatment of phase boundaries, map edges, and unindexed pixels;\\n- degrees versus radians;\\n- whether the map has been denoised before calculation.\\n\\nUse orix for disorientation and NumPy/SciPy for neighborhood operations. Store the number of valid neighbors along with KAM; otherwise values near pores, Si particles, and map boundaries are misleading. KAM should not be interpreted directly as dislocation density.\\n\\n### GND density\\n\\nGND is the least turnkey part of a MATLAB-free pipeline. A defensible calculation needs orientation gradients in physical units, lattice curvature, an explicit Nye-tensor convention, Burgers-vector/slip-system definitions, and a solution to an underdetermined inversion\\u2014commonly L1 or L2 minimization. Report whether the result is total scalar GND density, a lower bound, or slip-system-resolved density.\\n\\nFor conventional EBSD, noise and angular resolution strongly affect spatial derivatives. The result also scales with step size and preprocessing. Therefore:\\n\\n1. preserve orientations at full precision;\\n2. estimate angular noise on an undeformed/reference region;\\n3. document derivative stencil and smoothing;\\n4. perform step-size or downsampling sensitivity tests;\\n5. benchmark synthetic rotation-gradient fields with known curvature;\\n6. compare a representative map against an established MTEX or HR-EBSD implementation.\\n\\nNo named Python package in this comparison offers a broadly validated, turnkey GND workflow matching MTEX\\u2019s integration. Treat GND as an independently versioned scientific module, not as a plotting option attached to the grain segmenter.\\n\\n## Practical ranking\\n\\n- **Best production grain segmentation and statistics:** DREAM3D-NX/`simplnx`.\\n- **Best Python crystallographic foundation and PF/IPF layer:** orix.\\n- **Best upstream pattern/indexing integration:** kikuchipy + PyEBSDIndex.\\n- **Best lightweight 2D grain/deformation package:** DefDAP.\\n- **Best for TEM diffraction-microscopy integration, not SEM-EBSD post-processing:** pyxem.\\n- **Best custom microstructure/simulation framework:** PyMicro.\\n- **No fully satisfactory turnkey Python winner for GND:** implement and validate separately.\\n\\nAccordingly, the recommended deployable stack is **kikuchipy/PyEBSDIndex \\u2192 DREAM3D-NX/`simplnx` \\u2192 orix-based texture and KAM \\u2192 validated custom GND module**. This removes MATLAB from routine production while retaining MTEX only as an optional one-time validation reference. DREAM.3D\\u2019s documented use on complex additively manufactured 3D EBSD datasets, including automated parameter studies and explicit misorientation segmentation, makes this architecture substantially lower-risk than building the entire downstream stage directly from NumPy/orix primitives. (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000010, pqac-00000011)\"}", + "name": "answer" + } + } + ], + "info": null + } + ], + "hide_old_env_states": false, + "hide_old_action_content": false, + "sliding_window": null + }, + "observation": [ + { + "role": "tool", + "content": "Updated plan:\n{\n \"objectives\": [\n {\n \"id\": 1,\n \"objective\": \"Define evaluation criteria and identify candidate open-source Python-native EBSD post-processing packages relevant to grain reconstruction, texture, KAM, and GND.\",\n \"rationale\": \"The question asks for a comparative software landscape tailored to a MATLAB-free automated pipeline.\",\n \"status\": \"completed\",\n \"result\": \"Evaluation dimensions established: Python-native/API status, orientation-map ingestion, symmetry-aware grain segmentation, grain statistics, ODF/PF/IPF, KAM/GROD, GND/curvature, automation, maturity and peer-reviewed validation. Candidates: kikuchipy+orix, DefDAP, DREAM3D-NX/simplnx, pyxem, PyMicro, OpenXY/pyEBSD-style research code, and general scientific Python building blocks.\",\n \"evaluation\": \"Covers all requested functions and named packages while distinguishing integrated suites from libraries and research code.\"\n },\n {\n \"id\": 2,\n \"objective\": \"Gather primary and peer-reviewed evidence for kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and additional candidates.\",\n \"rationale\": \"Claims about capabilities, maturity, interfaces, and references must be grounded in software papers and official technical literature.\",\n \"status\": \"completed\",\n \"result\": \"Retrieved peer-reviewed evidence for DREAM.3D's mature pipeline-based 2D/3D EBSD reconstruction and segmentation, plus bibliographic records for orix (Johnstone et al., JAC 2020), kikuchipy, DefDAP (Zenodo software release), and related ecosystems. Evidence availability is strongest for DREAM.3D and MTEX; several package papers were indexed but full text unobtainable, so capability details must be presented conservatively and distinguish documented APIs from peer-reviewed validation.\",\n \"evaluation\": \"Enough evidence exists for a defensible package comparison, with explicit caveats where the scholarly record or retrieved full text is thin.\"\n },\n {\n \"id\": 3,\n \"objective\": \"Establish MTEX's benchmark capabilities for grain reconstruction, texture analysis, KAM, and GND to enable fair comparison.\",\n \"rationale\": \"The user explicitly requests comparisons with MTEX, so its relevant functionality must be documented.\",\n \"status\": \"completed\",\n \"result\": \"MTEX is the benchmark integrated environment: established symmetry-aware 2D grain detection, rich grain/boundary objects and specialized reconstruction extensions; mature ODF/PF/IPF and texture-component analysis; KAM/GROD and curvature-to-GND workflows. Peer-reviewed examples document advanced twin-hierarchy reconstruction and ODF/PF/IPF use. Threshold sensitivity and low-angle boundaries remain methodological issues rather than package-specific defects.\",\n \"evaluation\": \"Provides the functional baseline required to assess whether Python alternatives are replacements or complementary components.\"\n },\n {\n \"id\": 4,\n \"objective\": \"Synthesize a practical recommendation and workflow for downstream characterization of kikuchipy/PyEBSDIndex orientation maps from AM AlSi10Mg without MATLAB.\",\n \"rationale\": \"The answer should support a concrete implementation decision, not merely list software.\",\n \"status\": \"completed\",\n \"result\": \"Recommended architecture: retain kikuchipy/PyEBSDIndex for patterns/indexing; normalize map data into orix and/or DREAM3D-compatible arrays; use DREAM3D-NX/simplnx for robust repeatable grain segmentation/statistics, or DefDAP for lighter 2D Python grain/deformation analysis; use orix for symmetry-aware orientation operations and PF/IPF visualization; implement/validate ODF KDE and KAM explicitly; treat GND as a separate validated curvature/dislocation inversion module. Cross-check a subset against MTEX or certified reference output and preserve masks, phase IDs, step size, reference frames, and thresholds.\",\n \"evaluation\": \"The workflow directly addresses automated AM AlSi10Mg processing and honestly separates turnkey capabilities from custom work.\"\n },\n {\n \"id\": 5,\n \"objective\": \"Create a concise comparison artifact summarizing package status, capabilities, gaps, and best uses.\",\n \"rationale\": \"A structured matrix will make the multi-package comparison actionable.\",\n \"status\": \"completed\",\n \"result\": \"Created artifact-00, a package-by-package matrix covering Python status, grains, texture, KAM/GND, maturity, and recommended role.\",\n \"evaluation\": \"The matrix directly supports package selection and clearly distinguishes full, partial, and non-core capabilities.\"\n },\n {\n \"id\": 6,\n \"objective\": \"Call the answer tool with a thorough, evidence-cited technical response.\",\n \"rationale\": \"This is required to submit the final answer after all research objectives are complete.\",\n \"status\": \"in-progress\",\n \"result\": null,\n \"evaluation\": null\n }\n ]\n}", + "name": "update_plan", + "tool_call_id": "call_0Ow8AsXKD75JrGG2nzDUQk6Q", + "info": { + "start_ts": 1037.185396381, + "end_ts": 1037.186815636 + } + } + ], + "next_observation": [ + { + "role": "tool", + "content": "Answer was successfully submitted. Plan is not complete. Terminating the run anyway.", + "name": "answer", + "tool_call_id": "call_8LdlgnIRhAHn08c2v67pYdEC", + "info": { + "start_ts": 1127.880681022, + "end_ts": 1128.33717813 + } + } + ], + "reward": 0.0, + "truncated": false, + "done": true, + "value": 0.0, + "metadata": { + "time_elapsed_before_transition": 1.2371912090000023, + "time_elapsed_agent_get_asv": 71.43653710299986, + "time_elapsed_after_agent_get_asv": 1.0605452110000897, + "time_elapsed_env_step": 2.5720698879999873, + "time_elapsed_after_env_step": 1.0745023239999227 + }, + "action": { + "call_id": { + "run_id": "1e46df3a-e7dd-43f1-824c-21db05262b1a", + "fwd_id": "63e8080f-282c-44f6-acc9-fdadeb6f1766" + }, + "op_name": "IdentityOp-675644dba629", + "op_class_name": "ldp.graph.common_ops.IdentityOp", + "value": { + "role": "assistant", + "content": null, + "function_call": null, + "tool_calls": [ + { + "id": "call_8LdlgnIRhAHn08c2v67pYdEC", + "type": "function", + "function": { + "arguments": "{\"answer\": \"# Open-source Python options for EBSD post-processing\\n\\n## Executive conclusion\\n\\nThere is currently **no single Python-native package that reproduces the full integrated scope and maturity of MTEX**. The most credible MATLAB-free solution is a composed stack:\\n\\n1. **kikuchipy + PyEBSDIndex** for pattern processing and indexing;\\n2. **DREAM3D-NX through `simplnx`** for reproducible grain segmentation, cleanup, grain statistics, and especially 3D-capable workflows;\\n3. **orix** as the crystallographic orientation, symmetry, misorientation, IPF, and pole-figure layer;\\n4. optionally **DefDAP** for convenient 2D grain/deformation-map analysis;\\n5. a separately tested module for **KAM**, and especially for **GND-density inversion**.\\n\\nFor the proposed automated AlSi10Mg pipeline, DREAM3D-NX/`simplnx` is the strongest grain-characterization backend, while orix is the best Python-native crystallographic foundation. kikuchipy itself is not a downstream replacement for MTEX, and pyxem is primarily a diffraction-microscopy/4D-STEM package rather than a conventional SEM-EBSD grain-analysis suite.\\n\\n\\n\\n## The MTEX benchmark\\n\\nMTEX combines, in one environment, symmetry-aware EBSD import and manipulation, grain reconstruction, grain and boundary objects, texture estimation and plotting, misorientation statistics, KAM/GROD-type measures, and curvature/dislocation-density workflows. Its texture functionality includes ODF estimation, pole and inverse-pole figures, texture components, fibres, and quantitative texture measures. Published applications use MTEX for IPF maps, ODF-derived pole figures, grain-size and grain-shape statistics, and grain-boundary analysis. (pqac-00000000)\\n\\nMTEX also supports sophisticated extensions beyond simple threshold segmentation. A peer-reviewed workflow built on MTEX and MATLAB graph tools reconstructed deformation-twin hierarchies, grouped deformed fragments into parent families, treated missing relationships, and identified third-generation twins in heavily twinned \\u03b1-Ti. (pqac-00000009, pqac-00000012) This illustrates the principal gap in Python: individual Python libraries provide many underlying operations, but MTEX supplies a more coherent grain/boundary/texture abstraction and a larger collection of specialized reconstruction workflows.\\n\\nMTEX is not immune to methodological choices. Strong textures and low-angle boundaries can cause under-segmentation or chains of crystallographically similar regions; one MTEX study deliberately biased reconstruction toward over-segmentation and used 1\\u20132\\u00b0 tolerances for its particular material. (pqac-00000013) More generally, grain counts and geometry depend strongly on the selected threshold, connectivity, cleanup, and minimum-grain-size rules. A conventional 5\\u00b0 threshold is common but is not universally valid. (pqac-00000008)\\n\\n## Package assessment\\n\\n### 1. kikuchipy\\n\\n**Role.** kikuchipy is the natural upstream package in the proposed workflow. It is designed around EBSD patterns: vendor-file access, lazy handling of large pattern arrays, detector and projection geometry, background correction, pattern processing, simulation-related operations, and indexing interoperability. Its data model uses the HyperSpy ecosystem and orix crystallographic objects.\\n\\n**Grains and texture.** It should not be selected as the principal grain-reconstruction engine. Once indexing has produced orientation, phase, quality, and coordinate arrays, downstream segmentation and grain statistics generally require another package or custom code. Likewise, kikuchipy is not a comprehensive ODF/texture-analysis system and does not offer an MTEX-equivalent grain/boundary object model.\\n\\n**KAM/GND.** Neither is a core turnkey kikuchipy capability. KAM can be calculated from its output arrays with orix/NumPy-style operations, but masking, symmetry reduction, neighborhood definition, and exclusion of high-angle boundaries remain the pipeline developer\\u2019s responsibility. GND inversion is substantially further outside its main scope.\\n\\n**Maturity.** kikuchipy is a well-developed, domain-specific open-source project with a software paper titled *kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns*. Its maturity is strongest for pattern-space operations, not for post-indexing microstructure analysis.\\n\\n### 2. orix\\n\\n**Role.** orix is the most important Python-native crystallographic foundation. It supplies rotations and orientations, crystal and specimen symmetry, fundamental-zone reduction, symmetry-aware misorientation, orientation colours, inverse-pole-figure colour keys, pole-figure-style projections, and orientation clustering. Its peer-reviewed methodological reference is Johnstone et al., *Density-based clustering of crystal (mis)orientations and the orix Python library*, *Journal of Applied Crystallography* 53 (2020), 1293\\u20131298, DOI 10.1107/S1600576720011103.\\n\\n**Grain reconstruction.** orix contains the primitives needed to compare neighboring orientations and cluster orientation data, but it is not yet an MTEX-like turnkey EBSD topology package. A production grain segmenter still needs scan-grid connectivity, phase masks, confidence masks, flood-fill or graph labeling, small-region handling, and grain-property aggregation. Those steps can be implemented with NumPy, SciPy, scikit-image, xarray, and networkx, but then the laboratory owns validation and maintenance.\\n\\n**Texture.** orix is strong for IPF colouring, stereographic projections, symmetry handling, and discrete orientation distributions. It is suitable for PF/IPF visualization and custom orientation-density estimation. Nevertheless, MTEX remains substantially richer and more established for quantitative ODF estimation, kernel-selection conventions, harmonic representations, fibres/components, uncertainty-aware texture workflows, and publication-ready texture analysis. An ODF reconstructed from discrete EBSD orientations is also parameter-dependent; published work notes sensitivity to ODF computation settings in MTEX itself, so Python/MTEX agreement requires matching kernels, bandwidths, symmetry, weights, and normalization. A modern EBSD texture study used pole figures to characterize weak texture and illustrates the need to retain a statistically representative orientation population. (pqac-00000006)\\n\\n**KAM/GND.** orix is a good basis for KAM because it provides symmetry-aware misorientation operations. It does not, by itself, turn KAM into a one-call validated EBSD workflow. GND requires spatial orientation gradients, reference-frame consistency, Burgers vectors/slip-system assumptions, Nye-tensor construction, and an underdetermined inversion; this remains custom research code rather than an MTEX-equivalent orix feature.\\n\\n### 3. DefDAP\\n\\n**Role.** DefDAP\\u2014Deformation Data Analysis in Python\\u2014is a Python package aimed at combining EBSD with deformation measurements such as high-resolution digital image correlation. Its citable software release is Atkinson et al., *DefDAP: Deformation Data Analysis in Python* (2020), DOI 10.5281/zenodo.3688097.\\n\\n**Grains.** DefDAP is useful for 2D EBSD maps, grain finding and grain-level analysis, crystallographic slip/twinning interpretation, boundary overlays, and linking deformation fields to grains. For a conventional single-phase 2D Al EBSD map it may be the quickest lightweight route from an orientation map to grains and grain-wise quantities.\\n\\n**Texture and misorientation metrics.** Its emphasis is local deformation and grain-scale mechanics, not comprehensive quantitative texture. It is less complete than MTEX for ODFs, texture components, and generalized boundary/reconstruction operations. KAM-like local misorientation calculations can be added or may be available through map-level operations depending on release, but GND-density inversion should not be assumed to be a complete, standardized capability without checking the exact version and validating the implementation.\\n\\n**Maturity.** DefDAP has been used in peer-reviewed deformation studies, but its community, documentation breadth, and EBSD feature coverage are smaller than MTEX or DREAM.3D. It is best regarded as a focused 2D deformation-analysis package rather than a universal EBSD backend.\\n\\n### 4. DREAM.3D and DREAM3D-NX/`simplnx`\\n\\n**Role and architecture.** Legacy DREAM.3D is an established C++/Qt microstructure-analysis platform. DREAM3D-NX is its current-generation successor; `simplnx` provides Python bindings and pipeline construction. Thus, it is **Python-accessible and scriptable**, although not pure Python internally. This distinction is normally beneficial for large EBSD maps because segmentation and feature-statistics filters execute in compiled code.\\n\\n**Grain reconstruction.** This is the strongest open-source alternative for automated grain segmentation. DREAM.3D uses pipeline filters for data import, masking, cleanup, symmetry-aware misorientation segmentation, feature identification, neighbor lists, grain morphology, crystallographic statistics, boundary meshes, and, where applicable, 3D reconstruction. Published work describes DREAM.3D as a unified data structure aggregating community-developed algorithms, with reconstruction pipelines commonly containing more than a dozen parameterized steps. (pqac-00000011)\\n\\nIts `Segment Features (Misorientation)` workflow has been used to compare 2D and 3D segmentation across several misorientation thresholds. (pqac-00000007) Published AM/TriBeam workflows reconstructed serial EBSD slices, segmented grains using explicit misorientation tolerances, removed features below specified voxel counts, filled gaps, and automated parameter sweeps. (pqac-00000003, pqac-00000010) Cloud integration further demonstrated parallel parameter studies, provenance tracking, and grain-size comparisons between 2D and 3D AM datasets. (pqac-00000001, pqac-00000002)\\n\\n**Texture.** DREAM.3D calculates crystallographic feature statistics and supports orientation-distribution data used in synthetic microstructure generation, but it is not as strong as MTEX for interactive quantitative texture analysis. PF/IPF maps and basic orientation statistics are feasible; sophisticated ODF reconstruction and texture-component analysis are better handled in orix plus custom numerical code\\u2014or validated against MTEX. Published workflows frequently use DREAM.3D for microstructure generation/reconstruction and MTEX for detailed ODF/PF analysis, reflecting this division of strengths. (pqac-00000000)\\n\\n**KAM/GND.** DREAM3D-NX should not be selected primarily for KAM or GND. Depending on installed plugins and version, local orientation filters may cover some neighborhood measures, but these capabilities are less central and less transparent than its feature segmentation/statistics pipeline. A NumPy/orix KAM stage is easier to audit. GND remains a specialist calculation.\\n\\n**Cautions.** Segmentation is not parameter-free. The same threshold can produce different 2D and 3D connectivity, especially for nonconvex grains or paths connected outside the selected plane. (pqac-00000005, pqac-00000007, pqac-00000008) Minimum feature size and cleanup can also shift the measured grain-size distribution. Consequently, pipeline JSON, threshold, neighborhood convention, phase mask, and deleted/fill operations should be retained as provenance.\\n\\n### 5. pyxem\\n\\npyxem is an open-source Python ecosystem for multidimensional diffraction microscopy, built around HyperSpy and commonly used for scanning/precession electron diffraction and 4D-STEM. It is valuable for virtual diffraction imaging, diffraction-vector analysis, orientation mapping, and related TEM workflows.\\n\\nIt is **not a direct downstream package for conventional SEM-EBSD orientation maps**. Grain segmentation from vendor-style EBSD grids, quantitative EBSD texture analysis, KAM, and GND are not its central abstractions. Its shared ecosystem with kikuchipy/orix can be useful if the project later combines EBSD with scanning electron diffraction, but pyxem should not be the primary AlSi10Mg grain-analysis backend.\\n\\n### 6. PyMicro\\n\\nPyMicro provides Python classes and algorithms for crystallography, microstructure representation, grains, orientations, meshes, and links to simulation and tomography. It can import or represent EBSD-like microstructures and is useful for custom data integration, synthetic microstructures, and computational-mechanics coupling.\\n\\nFor this task, its limitations are that conventional EBSD segmentation, comprehensive ODF analysis, KAM, and GND are not offered with the same turnkey breadth or community validation as MTEX. It is better viewed as an extensible microstructure framework than as a drop-in EBSD analysis application.\\n\\n### 7. OpenXY and other research code\\n\\nOpenXY is associated mainly with high-angular-resolution EBSD cross-correlation and strain/rotation analysis. It addresses a different problem from ordinary Hough-indexed orientation-map characterization and is not a complete grain/texture package. Some Python ports and research repositories calculate lattice curvature or GND density, but their conventions, slip-system bases, regularization, and maintenance vary.\\n\\nOther small packages carrying names such as `pyebsd`, custom scikit-image segmenters, or laboratory scripts can calculate IPF colours, KAM, or threshold-connected grains. They can be useful building blocks, but package names are sometimes ambiguous, releases may be unmaintained, and peer-reviewed software references are often absent. They should not be treated as equivalent to MTEX merely because they produce an IPF or KAM image.\\n\\n## Recommended pipeline for AM AlSi10Mg\\n\\n### Data contract after indexing\\n\\nExport or retain the following arrays and metadata:\\n\\n- quaternion or rotation matrix per scan point\\u2014preferably not only Euler angles;\\n- phase ID, indexed/unindexed mask, confidence or match score, and pattern quality;\\n- physical `x`, `y` coordinates and step size;\\n- scan shape and rectangular/hexagonal topology;\\n- crystal symmetry and lattice parameters for fcc Al;\\n- detector/sample/reference-frame conventions;\\n- indexing dictionary or phase-library version.\\n\\nAlSi10Mg is commonly dominated in EBSD by fcc \\u03b1-Al, while the fine Si network may be incompletely indexed at ordinary step sizes. Do not let unindexed Si-rich pixels automatically split Al grains: establish an explicit phase/mask and cleanup policy and report it.\\n\\n### Grain stage\\n\\nUse **DREAM3D-NX/`simplnx`** for the production pipeline:\\n\\n1. construct the image geometry and load quaternion, phase, mask, and quality arrays;\\n2. remove only demonstrably unreliable points;\\n3. segment same-phase connected pixels with a documented disorientation threshold;\\n4. calculate feature centroids, equivalent diameters, areas, aspect ratios, mean orientations, neighbors, and boundary misorientations;\\n5. retain both the raw feature labels and any post-cleanup labels;\\n6. sweep plausible thresholds rather than fixing one unexamined value.\\n\\nFor AM AlSi10Mg, evaluate at least a low-angle-sensitive threshold and a conventional grain threshold\\u2014for example, a controlled sweep around 2\\u20135\\u00b0\\u2014because cellular/subgrain orientation gradients may be physically meaningful. This is not a recommendation to choose a universal value; published work demonstrates that changing tolerance can radically alter segmentation and that 2D and 3D results are not automatically equivalent. (pqac-00000003, pqac-00000005, pqac-00000007)\\n\\nIf only modest 2D maps are involved and direct Python object manipulation is more important than DREAM3D-NX pipelines, **DefDAP** is a reasonable alternative. A fully custom orix/scikit-image segmenter is defensible only if regression-tested against DREAM3D-NX or MTEX.\\n\\n### Texture stage\\n\\nUse **orix** for orientation conversion, symmetry reduction, IPF maps, pole figures, and discrete orientation statistics. For ODF estimation:\\n\\n- weight points or grains deliberately; pixel-weighted and one-orientation-per-grain ODFs answer different questions;\\n- state the kernel and half-width/bandwidth;\\n- enforce cubic crystal symmetry and the intended specimen symmetry;\\n- normalize PF/ODF intensities consistently, usually in multiples of random distribution;\\n- validate against a known texture or a small MTEX reference dataset.\\n\\nThis distinction matters in AM material: pixel weighting emphasizes large columnar grains, whereas grain weighting describes the population of grains. A recent study highlights the importance of preserving the full orientation distribution when reducing EBSD orientation datasets and uses PF maps to assess texture strength. (pqac-00000006)\\n\\n### KAM\\n\\nKAM is straightforward to implement but easy to define inconsistently. For each indexed point, compute the mean symmetry-reduced disorientation to valid neighbors, while specifying:\\n\\n- first-, second-, or higher-order neighborhood;\\n- whether diagonals are included;\\n- maximum included disorientation, often chosen to exclude grain boundaries;\\n- treatment of phase boundaries, map edges, and unindexed pixels;\\n- degrees versus radians;\\n- whether the map has been denoised before calculation.\\n\\nUse orix for disorientation and NumPy/SciPy for neighborhood operations. Store the number of valid neighbors along with KAM; otherwise values near pores, Si particles, and map boundaries are misleading. KAM should not be interpreted directly as dislocation density.\\n\\n### GND density\\n\\nGND is the least turnkey part of a MATLAB-free pipeline. A defensible calculation needs orientation gradients in physical units, lattice curvature, an explicit Nye-tensor convention, Burgers-vector/slip-system definitions, and a solution to an underdetermined inversion\\u2014commonly L1 or L2 minimization. Report whether the result is total scalar GND density, a lower bound, or slip-system-resolved density.\\n\\nFor conventional EBSD, noise and angular resolution strongly affect spatial derivatives. The result also scales with step size and preprocessing. Therefore:\\n\\n1. preserve orientations at full precision;\\n2. estimate angular noise on an undeformed/reference region;\\n3. document derivative stencil and smoothing;\\n4. perform step-size or downsampling sensitivity tests;\\n5. benchmark synthetic rotation-gradient fields with known curvature;\\n6. compare a representative map against an established MTEX or HR-EBSD implementation.\\n\\nNo named Python package in this comparison offers a broadly validated, turnkey GND workflow matching MTEX\\u2019s integration. Treat GND as an independently versioned scientific module, not as a plotting option attached to the grain segmenter.\\n\\n## Practical ranking\\n\\n- **Best production grain segmentation and statistics:** DREAM3D-NX/`simplnx`.\\n- **Best Python crystallographic foundation and PF/IPF layer:** orix.\\n- **Best upstream pattern/indexing integration:** kikuchipy + PyEBSDIndex.\\n- **Best lightweight 2D grain/deformation package:** DefDAP.\\n- **Best for TEM diffraction-microscopy integration, not SEM-EBSD post-processing:** pyxem.\\n- **Best custom microstructure/simulation framework:** PyMicro.\\n- **No fully satisfactory turnkey Python winner for GND:** implement and validate separately.\\n\\nAccordingly, the recommended deployable stack is **kikuchipy/PyEBSDIndex \\u2192 DREAM3D-NX/`simplnx` \\u2192 orix-based texture and KAM \\u2192 validated custom GND module**. This removes MATLAB from routine production while retaining MTEX only as an optional one-time validation reference. DREAM.3D\\u2019s documented use on complex additively manufactured 3D EBSD datasets, including automated parameter studies and explicit misorientation segmentation, makes this architecture substantially lower-risk than building the entire downstream stage directly from NumPy/orix primitives. (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000010, pqac-00000011)\"}", + "name": "answer" + } + } + ] + }, + "logprob": null + } + } + }, + "created_at": "2026-07-29T03:41:08.424773Z" + } + ], + "environment_frame": { + "id": "0f3443d3-895c-4a6a-b22f-094719db1fba", + "agent_state_point_in_time": null, + "trajectory_timestep": 14, + "current_agent_step": "after_transition", + "trajectory_id": "f55d0bdd-87d8-4cf2-8d86-23309987262a", + "state": { + "deepcopy": true, + "state": { + "id": "f55d0bdd-87d8-4cf2-8d86-23309987262a", + "created_at": "2026-07-29T03:41:37.368542", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "request": {}, + "response": { + "answer": { + "id": "f55d0bdd-87d8-4cf2-8d86-23309987262a", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "answer": "# Open-source Python options for EBSD post-processing\n\n## Executive conclusion\n\nThere is currently **no single Python-native package that reproduces the full integrated scope and maturity of MTEX**. The most credible MATLAB-free solution is a composed stack:\n\n1. **kikuchipy + PyEBSDIndex** for pattern processing and indexing;\n2. **DREAM3D-NX through `simplnx`** for reproducible grain segmentation, cleanup, grain statistics, and especially 3D-capable workflows;\n3. **orix** as the crystallographic orientation, symmetry, misorientation, IPF, and pole-figure layer;\n4. optionally **DefDAP** for convenient 2D grain/deformation-map analysis;\n5. a separately tested module for **KAM**, and especially for **GND-density inversion**.\n\nFor the proposed automated AlSi10Mg pipeline, DREAM3D-NX/`simplnx` is the strongest grain-characterization backend, while orix is the best Python-native crystallographic foundation. kikuchipy itself is not a downstream replacement for MTEX, and pyxem is primarily a diffraction-microscopy/4D-STEM package rather than a conventional SEM-EBSD grain-analysis suite.\n\n| Package / ecosystem | Python-native status | Grain reconstruction / segmentation | Texture ODF / PF / IPF | KAM / GND | Maturity / reference status | Recommended role |\n|---|---|---|---|---|---|---|\n| MTEX | No; MATLAB-based | **Strong** for symmetry-aware 2D grain detection/reconstruction; advanced specialized workflows documented | **Strong**; benchmark for ODF, pole figures, IPF, texture components (folwarczny2026accurategrainboundary pages 5-9, savage2021anautomatedprocedure pages 1-2, savage2021anautomatedprocedure pages 3-4) | **Strong** in the MTEX ecosystem; widely treated as benchmark | Very mature, heavily cited benchmark; peer-reviewed advanced EBSD workflows documented (savage2021anautomatedprocedure pages 1-2, savage2021anautomatedprocedure pages 3-4) | Functional benchmark and validation target, but not MATLAB-free |\n| kikuchipy | Yes | **Absent / not core** for downstream grain reconstruction; primarily EBSD pattern processing/indexing support | **Partial / custom** via interoperability, not a full texture-analysis suite | **Absent / not core** | Open-source Python EBSD toolbox with literature record, but not a complete post-processing replacement in the retrieved evidence | Upstream pattern handling, preprocessing, indexing, data I/O; pair with orix and/or DREAM3D-NX |\n| orix | Yes | **Partial / custom**; orientation objects and misorientation logic support custom grain segmentation, but not MTEX-level turnkey reconstruction | **Strong / near-benchmark for orientation visualization**; IPF/PF native, ODF possible but typically less turnkey than MTEX | **Partial / custom**; KAM-like metrics and curvature workflows need user implementation/validation | Peer-reviewed library record available; positioned as crystallographic orientation library rather than full EBSD suite | Core Python orientation engine for custom EBSD post-processing and visualization |\n| DefDAP | Yes | **Partial**; useful 2D deformation/EBSD map grain analysis, but narrower than MTEX | **Partial**; some crystallographic plotting/analysis, not a full MTEX texture environment | **Partial**; deformation-oriented metrics are a strength, GND generally custom/limited | Open-source Python project with software-release citation; maturity moderate, community smaller than MTEX/DREAM3D | Good lightweight Python choice for 2D grain/deformation analysis after indexing |\n| DREAM3D-NX / simplnx | Yes (Python API around DREAM3D-NX) | **Strong**; mature pipeline-based segmentation/reconstruction, especially for 2D/3D EBSD and batch automation (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5, yi2025rapidgrainsegmentation pages 4-7, stanek2020comparisonofsegmentation pages 1-2) | **Partial**; some crystallographic statistics, but not as texture-centric or interactive as MTEX (yi2025rapidgrainsegmentation pages 4-7) | **Partial / custom**; not typically the first choice for KAM/GND-centric analysis | Mature EBSD/microstructure pipeline ecosystem with peer-reviewed reconstruction studies (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5) | Best Python-accessible option for robust automated grain segmentation/statistics in a MATLAB-free pipeline |\n| pyxem | Yes | **Absent / not core for SEM-EBSD**; focused on diffraction microscopy in TEM/4D-STEM, not a direct MTEX replacement | **Partial** for orientation mapping/visualization in diffraction-microscopy contexts | **Absent / not core** | Mature open-source microscopy library, but its center of gravity is 4D-STEM/SPED rather than conventional SEM-EBSD post-processing | Use only if your workflow spans diffraction microscopy; not recommended as primary EBSD grain-texture backend |\n| PyMicro | Yes | **Partial / custom**; useful microstructure data structures and crystallography, but less turnkey for EBSD grain reconstruction | **Partial / custom** | **Partial / custom** | Useful research software ecosystem, but less established as a mainstream EBSD post-processing stack in the retrieved evidence | Supporting library for custom workflows, data models, simulation coupling |\n| OpenXY / research scripts / pyEBSD-style code | Yes, but heterogeneous | **Partial / custom** to **strong in isolated tasks**; highly variable and often method-specific | **Partial / custom** | **Partial / custom**, including possible GND research implementations | Lowest standardization; often lacks broad peer-reviewed software validation or stable APIs | Fill specific gaps such as KAM/GND or custom segmentation, but validate carefully against MTEX/DREAM3D outputs |\n| Practical no-MATLAB stack for AM AlSi10Mg | Yes | **Strong overall** if composed as: kikuchipy/PyEBSDIndex + DREAM3D-NX for segmentation + orix/DefDAP for orientation/statistics | **Partial-to-strong overall**; strongest with orix plus custom ODF tooling | **Partial / custom overall**; KAM feasible, GND should be a separately validated module | Most realistic MATLAB-free path based on current open-source ecosystem and evidence balance (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5) | Recommended integrated pipeline rather than searching for a single Python package that fully replaces MTEX |\n\n\n*Table: This table compares major open-source Python-native or Python-accessible EBSD post-processing options against MTEX for grain reconstruction, texture analysis, and misorientation-derived metrics. It is designed to help choose a MATLAB-free downstream pipeline for automated EBSD workflows.*\n\n## The MTEX benchmark\n\nMTEX combines, in one environment, symmetry-aware EBSD import and manipulation, grain reconstruction, grain and boundary objects, texture estimation and plotting, misorientation statistics, KAM/GROD-type measures, and curvature/dislocation-density workflows. Its texture functionality includes ODF estimation, pole and inverse-pole figures, texture components, fibres, and quantitative texture measures. Published applications use MTEX for IPF maps, ODF-derived pole figures, grain-size and grain-shape statistics, and grain-boundary analysis. (folwarczny2026accurategrainboundary pages 5-9)\n\nMTEX also supports sophisticated extensions beyond simple threshold segmentation. A peer-reviewed workflow built on MTEX and MATLAB graph tools reconstructed deformation-twin hierarchies, grouped deformed fragments into parent families, treated missing relationships, and identified third-generation twins in heavily twinned \u03b1-Ti. (savage2021anautomatedprocedure pages 1-2) This illustrates the principal gap in Python: individual Python libraries provide many underlying operations, but MTEX supplies a more coherent grain/boundary/texture abstraction and a larger collection of specialized reconstruction workflows.\n\nMTEX is not immune to methodological choices. Strong textures and low-angle boundaries can cause under-segmentation or chains of crystallographically similar regions; one MTEX study deliberately biased reconstruction toward over-segmentation and used 1\u20132\u00b0 tolerances for its particular material. (savage2021anautomatedprocedure pages 3-4) More generally, grain counts and geometry depend strongly on the selected threshold, connectivity, cleanup, and minimum-grain-size rules. A conventional 5\u00b0 threshold is common but is not universally valid. (stanek2020comparisonofsegmentation pages 1-2)\n\n## Package assessment\n\n### 1. kikuchipy\n\n**Role.** kikuchipy is the natural upstream package in the proposed workflow. It is designed around EBSD patterns: vendor-file access, lazy handling of large pattern arrays, detector and projection geometry, background correction, pattern processing, simulation-related operations, and indexing interoperability. Its data model uses the HyperSpy ecosystem and orix crystallographic objects.\n\n**Grains and texture.** It should not be selected as the principal grain-reconstruction engine. Once indexing has produced orientation, phase, quality, and coordinate arrays, downstream segmentation and grain statistics generally require another package or custom code. Likewise, kikuchipy is not a comprehensive ODF/texture-analysis system and does not offer an MTEX-equivalent grain/boundary object model.\n\n**KAM/GND.** Neither is a core turnkey kikuchipy capability. KAM can be calculated from its output arrays with orix/NumPy-style operations, but masking, symmetry reduction, neighborhood definition, and exclusion of high-angle boundaries remain the pipeline developer\u2019s responsibility. GND inversion is substantially further outside its main scope.\n\n**Maturity.** kikuchipy is a well-developed, domain-specific open-source project with a software paper titled *kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns*. Its maturity is strongest for pattern-space operations, not for post-indexing microstructure analysis.\n\n### 2. orix\n\n**Role.** orix is the most important Python-native crystallographic foundation. It supplies rotations and orientations, crystal and specimen symmetry, fundamental-zone reduction, symmetry-aware misorientation, orientation colours, inverse-pole-figure colour keys, pole-figure-style projections, and orientation clustering. Its peer-reviewed methodological reference is Johnstone et al., *Density-based clustering of crystal (mis)orientations and the orix Python library*, *Journal of Applied Crystallography* 53 (2020), 1293\u20131298, DOI 10.1107/S1600576720011103.\n\n**Grain reconstruction.** orix contains the primitives needed to compare neighboring orientations and cluster orientation data, but it is not yet an MTEX-like turnkey EBSD topology package. A production grain segmenter still needs scan-grid connectivity, phase masks, confidence masks, flood-fill or graph labeling, small-region handling, and grain-property aggregation. Those steps can be implemented with NumPy, SciPy, scikit-image, xarray, and networkx, but then the laboratory owns validation and maintenance.\n\n**Texture.** orix is strong for IPF colouring, stereographic projections, symmetry handling, and discrete orientation distributions. It is suitable for PF/IPF visualization and custom orientation-density estimation. Nevertheless, MTEX remains substantially richer and more established for quantitative ODF estimation, kernel-selection conventions, harmonic representations, fibres/components, uncertainty-aware texture workflows, and publication-ready texture analysis. An ODF reconstructed from discrete EBSD orientations is also parameter-dependent; published work notes sensitivity to ODF computation settings in MTEX itself, so Python/MTEX agreement requires matching kernels, bandwidths, symmetry, weights, and normalization. A modern EBSD texture study used pole figures to characterize weak texture and illustrates the need to retain a statistically representative orientation population. (wanni2024machinelearningenhanced pages 1-2)\n\n**KAM/GND.** orix is a good basis for KAM because it provides symmetry-aware misorientation operations. It does not, by itself, turn KAM into a one-call validated EBSD workflow. GND requires spatial orientation gradients, reference-frame consistency, Burgers vectors/slip-system assumptions, Nye-tensor construction, and an underdetermined inversion; this remains custom research code rather than an MTEX-equivalent orix feature.\n\n### 3. DefDAP\n\n**Role.** DefDAP\u2014Deformation Data Analysis in Python\u2014is a Python package aimed at combining EBSD with deformation measurements such as high-resolution digital image correlation. Its citable software release is Atkinson et al., *DefDAP: Deformation Data Analysis in Python* (2020), DOI 10.5281/zenodo.3688097.\n\n**Grains.** DefDAP is useful for 2D EBSD maps, grain finding and grain-level analysis, crystallographic slip/twinning interpretation, boundary overlays, and linking deformation fields to grains. For a conventional single-phase 2D Al EBSD map it may be the quickest lightweight route from an orientation map to grains and grain-wise quantities.\n\n**Texture and misorientation metrics.** Its emphasis is local deformation and grain-scale mechanics, not comprehensive quantitative texture. It is less complete than MTEX for ODFs, texture components, and generalized boundary/reconstruction operations. KAM-like local misorientation calculations can be added or may be available through map-level operations depending on release, but GND-density inversion should not be assumed to be a complete, standardized capability without checking the exact version and validating the implementation.\n\n**Maturity.** DefDAP has been used in peer-reviewed deformation studies, but its community, documentation breadth, and EBSD feature coverage are smaller than MTEX or DREAM.3D. It is best regarded as a focused 2D deformation-analysis package rather than a universal EBSD backend.\n\n### 4. DREAM.3D and DREAM3D-NX/`simplnx`\n\n**Role and architecture.** Legacy DREAM.3D is an established C++/Qt microstructure-analysis platform. DREAM3D-NX is its current-generation successor; `simplnx` provides Python bindings and pipeline construction. Thus, it is **Python-accessible and scriptable**, although not pure Python internally. This distinction is normally beneficial for large EBSD maps because segmentation and feature-statistics filters execute in compiled code.\n\n**Grain reconstruction.** This is the strongest open-source alternative for automated grain segmentation. DREAM.3D uses pipeline filters for data import, masking, cleanup, symmetry-aware misorientation segmentation, feature identification, neighbor lists, grain morphology, crystallographic statistics, boundary meshes, and, where applicable, 3D reconstruction. Published work describes DREAM.3D as a unified data structure aggregating community-developed algorithms, with reconstruction pipelines commonly containing more than a dozen parameterized steps. (polonsky2019threedimensionalanalysisand pages 1-2)\n\nIts `Segment Features (Misorientation)` workflow has been used to compare 2D and 3D segmentation across several misorientation thresholds. (stanek2020comparisonofsegmentation pages 2-7) Published AM/TriBeam workflows reconstructed serial EBSD slices, segmented grains using explicit misorientation tolerances, removed features below specified voxel counts, filled gaps, and automated parameter sweeps. (polonsky2019threedimensionalanalysisand pages 2-5) Cloud integration further demonstrated parallel parameter studies, provenance tracking, and grain-size comparisons between 2D and 3D AM datasets. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14)\n\n**Texture.** DREAM.3D calculates crystallographic feature statistics and supports orientation-distribution data used in synthetic microstructure generation, but it is not as strong as MTEX for interactive quantitative texture analysis. PF/IPF maps and basic orientation statistics are feasible; sophisticated ODF reconstruction and texture-component analysis are better handled in orix plus custom numerical code\u2014or validated against MTEX. Published workflows frequently use DREAM.3D for microstructure generation/reconstruction and MTEX for detailed ODF/PF analysis, reflecting this division of strengths. (folwarczny2026accurategrainboundary pages 5-9)\n\n**KAM/GND.** DREAM3D-NX should not be selected primarily for KAM or GND. Depending on installed plugins and version, local orientation filters may cover some neighborhood measures, but these capabilities are less central and less transparent than its feature segmentation/statistics pipeline. A NumPy/orix KAM stage is easier to audit. GND remains a specialist calculation.\n\n**Cautions.** Segmentation is not parameter-free. The same threshold can produce different 2D and 3D connectivity, especially for nonconvex grains or paths connected outside the selected plane. (stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 2-7) Minimum feature size and cleanup can also shift the measured grain-size distribution. Consequently, pipeline JSON, threshold, neighborhood convention, phase mask, and deleted/fill operations should be retained as provenance.\n\n### 5. pyxem\n\npyxem is an open-source Python ecosystem for multidimensional diffraction microscopy, built around HyperSpy and commonly used for scanning/precession electron diffraction and 4D-STEM. It is valuable for virtual diffraction imaging, diffraction-vector analysis, orientation mapping, and related TEM workflows.\n\nIt is **not a direct downstream package for conventional SEM-EBSD orientation maps**. Grain segmentation from vendor-style EBSD grids, quantitative EBSD texture analysis, KAM, and GND are not its central abstractions. Its shared ecosystem with kikuchipy/orix can be useful if the project later combines EBSD with scanning electron diffraction, but pyxem should not be the primary AlSi10Mg grain-analysis backend.\n\n### 6. PyMicro\n\nPyMicro provides Python classes and algorithms for crystallography, microstructure representation, grains, orientations, meshes, and links to simulation and tomography. It can import or represent EBSD-like microstructures and is useful for custom data integration, synthetic microstructures, and computational-mechanics coupling.\n\nFor this task, its limitations are that conventional EBSD segmentation, comprehensive ODF analysis, KAM, and GND are not offered with the same turnkey breadth or community validation as MTEX. It is better viewed as an extensible microstructure framework than as a drop-in EBSD analysis application.\n\n### 7. OpenXY and other research code\n\nOpenXY is associated mainly with high-angular-resolution EBSD cross-correlation and strain/rotation analysis. It addresses a different problem from ordinary Hough-indexed orientation-map characterization and is not a complete grain/texture package. Some Python ports and research repositories calculate lattice curvature or GND density, but their conventions, slip-system bases, regularization, and maintenance vary.\n\nOther small packages carrying names such as `pyebsd`, custom scikit-image segmenters, or laboratory scripts can calculate IPF colours, KAM, or threshold-connected grains. They can be useful building blocks, but package names are sometimes ambiguous, releases may be unmaintained, and peer-reviewed software references are often absent. They should not be treated as equivalent to MTEX merely because they produce an IPF or KAM image.\n\n## Recommended pipeline for AM AlSi10Mg\n\n### Data contract after indexing\n\nExport or retain the following arrays and metadata:\n\n- quaternion or rotation matrix per scan point\u2014preferably not only Euler angles;\n- phase ID, indexed/unindexed mask, confidence or match score, and pattern quality;\n- physical `x`, `y` coordinates and step size;\n- scan shape and rectangular/hexagonal topology;\n- crystal symmetry and lattice parameters for fcc Al;\n- detector/sample/reference-frame conventions;\n- indexing dictionary or phase-library version.\n\nAlSi10Mg is commonly dominated in EBSD by fcc \u03b1-Al, while the fine Si network may be incompletely indexed at ordinary step sizes. Do not let unindexed Si-rich pixels automatically split Al grains: establish an explicit phase/mask and cleanup policy and report it.\n\n### Grain stage\n\nUse **DREAM3D-NX/`simplnx`** for the production pipeline:\n\n1. construct the image geometry and load quaternion, phase, mask, and quality arrays;\n2. remove only demonstrably unreliable points;\n3. segment same-phase connected pixels with a documented disorientation threshold;\n4. calculate feature centroids, equivalent diameters, areas, aspect ratios, mean orientations, neighbors, and boundary misorientations;\n5. retain both the raw feature labels and any post-cleanup labels;\n6. sweep plausible thresholds rather than fixing one unexamined value.\n\nFor AM AlSi10Mg, evaluate at least a low-angle-sensitive threshold and a conventional grain threshold\u2014for example, a controlled sweep around 2\u20135\u00b0\u2014because cellular/subgrain orientation gradients may be physically meaningful. This is not a recommendation to choose a universal value; published work demonstrates that changing tolerance can radically alter segmentation and that 2D and 3D results are not automatically equivalent. (polonsky2019threedimensionalanalysisand pages 2-5, stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 2-7)\n\nIf only modest 2D maps are involved and direct Python object manipulation is more important than DREAM3D-NX pipelines, **DefDAP** is a reasonable alternative. A fully custom orix/scikit-image segmenter is defensible only if regression-tested against DREAM3D-NX or MTEX.\n\n### Texture stage\n\nUse **orix** for orientation conversion, symmetry reduction, IPF maps, pole figures, and discrete orientation statistics. For ODF estimation:\n\n- weight points or grains deliberately; pixel-weighted and one-orientation-per-grain ODFs answer different questions;\n- state the kernel and half-width/bandwidth;\n- enforce cubic crystal symmetry and the intended specimen symmetry;\n- normalize PF/ODF intensities consistently, usually in multiples of random distribution;\n- validate against a known texture or a small MTEX reference dataset.\n\nThis distinction matters in AM material: pixel weighting emphasizes large columnar grains, whereas grain weighting describes the population of grains. A recent study highlights the importance of preserving the full orientation distribution when reducing EBSD orientation datasets and uses PF maps to assess texture strength. (wanni2024machinelearningenhanced pages 1-2)\n\n### KAM\n\nKAM is straightforward to implement but easy to define inconsistently. For each indexed point, compute the mean symmetry-reduced disorientation to valid neighbors, while specifying:\n\n- first-, second-, or higher-order neighborhood;\n- whether diagonals are included;\n- maximum included disorientation, often chosen to exclude grain boundaries;\n- treatment of phase boundaries, map edges, and unindexed pixels;\n- degrees versus radians;\n- whether the map has been denoised before calculation.\n\nUse orix for disorientation and NumPy/SciPy for neighborhood operations. Store the number of valid neighbors along with KAM; otherwise values near pores, Si particles, and map boundaries are misleading. KAM should not be interpreted directly as dislocation density.\n\n### GND density\n\nGND is the least turnkey part of a MATLAB-free pipeline. A defensible calculation needs orientation gradients in physical units, lattice curvature, an explicit Nye-tensor convention, Burgers-vector/slip-system definitions, and a solution to an underdetermined inversion\u2014commonly L1 or L2 minimization. Report whether the result is total scalar GND density, a lower bound, or slip-system-resolved density.\n\nFor conventional EBSD, noise and angular resolution strongly affect spatial derivatives. The result also scales with step size and preprocessing. Therefore:\n\n1. preserve orientations at full precision;\n2. estimate angular noise on an undeformed/reference region;\n3. document derivative stencil and smoothing;\n4. perform step-size or downsampling sensitivity tests;\n5. benchmark synthetic rotation-gradient fields with known curvature;\n6. compare a representative map against an established MTEX or HR-EBSD implementation.\n\nNo named Python package in this comparison offers a broadly validated, turnkey GND workflow matching MTEX\u2019s integration. Treat GND as an independently versioned scientific module, not as a plotting option attached to the grain segmenter.\n\n## Practical ranking\n\n- **Best production grain segmentation and statistics:** DREAM3D-NX/`simplnx`.\n- **Best Python crystallographic foundation and PF/IPF layer:** orix.\n- **Best upstream pattern/indexing integration:** kikuchipy + PyEBSDIndex.\n- **Best lightweight 2D grain/deformation package:** DefDAP.\n- **Best for TEM diffraction-microscopy integration, not SEM-EBSD post-processing:** pyxem.\n- **Best custom microstructure/simulation framework:** PyMicro.\n- **No fully satisfactory turnkey Python winner for GND:** implement and validate separately.\n\nAccordingly, the recommended deployable stack is **kikuchipy/PyEBSDIndex \u2192 DREAM3D-NX/`simplnx` \u2192 orix-based texture and KAM \u2192 validated custom GND module**. This removes MATLAB from routine production while retaining MTEX only as an optional one-time validation reference. DREAM.3D\u2019s documented use on complex additively manufactured 3D EBSD datasets, including automated parameter studies and explicit misorientation segmentation, makes this architecture substantially lower-risk than building the entire downstream stage directly from NumPy/orix primitives. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5)", + "raw_answer": "# Open-source Python options for EBSD post-processing\n\n## Executive conclusion\n\nThere is currently **no single Python-native package that reproduces the full integrated scope and maturity of MTEX**. The most credible MATLAB-free solution is a composed stack:\n\n1. **kikuchipy + PyEBSDIndex** for pattern processing and indexing;\n2. **DREAM3D-NX through `simplnx`** for reproducible grain segmentation, cleanup, grain statistics, and especially 3D-capable workflows;\n3. **orix** as the crystallographic orientation, symmetry, misorientation, IPF, and pole-figure layer;\n4. optionally **DefDAP** for convenient 2D grain/deformation-map analysis;\n5. a separately tested module for **KAM**, and especially for **GND-density inversion**.\n\nFor the proposed automated AlSi10Mg pipeline, DREAM3D-NX/`simplnx` is the strongest grain-characterization backend, while orix is the best Python-native crystallographic foundation. kikuchipy itself is not a downstream replacement for MTEX, and pyxem is primarily a diffraction-microscopy/4D-STEM package rather than a conventional SEM-EBSD grain-analysis suite.\n\n| Package / ecosystem | Python-native status | Grain reconstruction / segmentation | Texture ODF / PF / IPF | KAM / GND | Maturity / reference status | Recommended role |\n|---|---|---|---|---|---|---|\n| MTEX | No; MATLAB-based | **Strong** for symmetry-aware 2D grain detection/reconstruction; advanced specialized workflows documented | **Strong**; benchmark for ODF, pole figures, IPF, texture components (pqac-00000000, pqac-00000012, pqac-00000013) | **Strong** in the MTEX ecosystem; widely treated as benchmark | Very mature, heavily cited benchmark; peer-reviewed advanced EBSD workflows documented (pqac-00000012, pqac-00000013) | Functional benchmark and validation target, but not MATLAB-free |\n| kikuchipy | Yes | **Absent / not core** for downstream grain reconstruction; primarily EBSD pattern processing/indexing support | **Partial / custom** via interoperability, not a full texture-analysis suite | **Absent / not core** | Open-source Python EBSD toolbox with literature record, but not a complete post-processing replacement in the retrieved evidence | Upstream pattern handling, preprocessing, indexing, data I/O; pair with orix and/or DREAM3D-NX |\n| orix | Yes | **Partial / custom**; orientation objects and misorientation logic support custom grain segmentation, but not MTEX-level turnkey reconstruction | **Strong / near-benchmark for orientation visualization**; IPF/PF native, ODF possible but typically less turnkey than MTEX | **Partial / custom**; KAM-like metrics and curvature workflows need user implementation/validation | Peer-reviewed library record available; positioned as crystallographic orientation library rather than full EBSD suite | Core Python orientation engine for custom EBSD post-processing and visualization |\n| DefDAP | Yes | **Partial**; useful 2D deformation/EBSD map grain analysis, but narrower than MTEX | **Partial**; some crystallographic plotting/analysis, not a full MTEX texture environment | **Partial**; deformation-oriented metrics are a strength, GND generally custom/limited | Open-source Python project with software-release citation; maturity moderate, community smaller than MTEX/DREAM3D | Good lightweight Python choice for 2D grain/deformation analysis after indexing |\n| DREAM3D-NX / simplnx | Yes (Python API around DREAM3D-NX) | **Strong**; mature pipeline-based segmentation/reconstruction, especially for 2D/3D EBSD and batch automation (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000011) | **Partial**; some crystallographic statistics, but not as texture-centric or interactive as MTEX (pqac-00000004) | **Partial / custom**; not typically the first choice for KAM/GND-centric analysis | Mature EBSD/microstructure pipeline ecosystem with peer-reviewed reconstruction studies (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Best Python-accessible option for robust automated grain segmentation/statistics in a MATLAB-free pipeline |\n| pyxem | Yes | **Absent / not core for SEM-EBSD**; focused on diffraction microscopy in TEM/4D-STEM, not a direct MTEX replacement | **Partial** for orientation mapping/visualization in diffraction-microscopy contexts | **Absent / not core** | Mature open-source microscopy library, but its center of gravity is 4D-STEM/SPED rather than conventional SEM-EBSD post-processing | Use only if your workflow spans diffraction microscopy; not recommended as primary EBSD grain-texture backend |\n| PyMicro | Yes | **Partial / custom**; useful microstructure data structures and crystallography, but less turnkey for EBSD grain reconstruction | **Partial / custom** | **Partial / custom** | Useful research software ecosystem, but less established as a mainstream EBSD post-processing stack in the retrieved evidence | Supporting library for custom workflows, data models, simulation coupling |\n| OpenXY / research scripts / pyEBSD-style code | Yes, but heterogeneous | **Partial / custom** to **strong in isolated tasks**; highly variable and often method-specific | **Partial / custom** | **Partial / custom**, including possible GND research implementations | Lowest standardization; often lacks broad peer-reviewed software validation or stable APIs | Fill specific gaps such as KAM/GND or custom segmentation, but validate carefully against MTEX/DREAM3D outputs |\n| Practical no-MATLAB stack for AM AlSi10Mg | Yes | **Strong overall** if composed as: kikuchipy/PyEBSDIndex + DREAM3D-NX for segmentation + orix/DefDAP for orientation/statistics | **Partial-to-strong overall**; strongest with orix plus custom ODF tooling | **Partial / custom overall**; KAM feasible, GND should be a separately validated module | Most realistic MATLAB-free path based on current open-source ecosystem and evidence balance (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Recommended integrated pipeline rather than searching for a single Python package that fully replaces MTEX |\n\n\n*Table: This table compares major open-source Python-native or Python-accessible EBSD post-processing options against MTEX for grain reconstruction, texture analysis, and misorientation-derived metrics. It is designed to help choose a MATLAB-free downstream pipeline for automated EBSD workflows.*\n\n## The MTEX benchmark\n\nMTEX combines, in one environment, symmetry-aware EBSD import and manipulation, grain reconstruction, grain and boundary objects, texture estimation and plotting, misorientation statistics, KAM/GROD-type measures, and curvature/dislocation-density workflows. Its texture functionality includes ODF estimation, pole and inverse-pole figures, texture components, fibres, and quantitative texture measures. Published applications use MTEX for IPF maps, ODF-derived pole figures, grain-size and grain-shape statistics, and grain-boundary analysis. (pqac-00000000)\n\nMTEX also supports sophisticated extensions beyond simple threshold segmentation. A peer-reviewed workflow built on MTEX and MATLAB graph tools reconstructed deformation-twin hierarchies, grouped deformed fragments into parent families, treated missing relationships, and identified third-generation twins in heavily twinned \u03b1-Ti. (pqac-00000009, pqac-00000012) This illustrates the principal gap in Python: individual Python libraries provide many underlying operations, but MTEX supplies a more coherent grain/boundary/texture abstraction and a larger collection of specialized reconstruction workflows.\n\nMTEX is not immune to methodological choices. Strong textures and low-angle boundaries can cause under-segmentation or chains of crystallographically similar regions; one MTEX study deliberately biased reconstruction toward over-segmentation and used 1\u20132\u00b0 tolerances for its particular material. (pqac-00000013) More generally, grain counts and geometry depend strongly on the selected threshold, connectivity, cleanup, and minimum-grain-size rules. A conventional 5\u00b0 threshold is common but is not universally valid. (pqac-00000008)\n\n## Package assessment\n\n### 1. kikuchipy\n\n**Role.** kikuchipy is the natural upstream package in the proposed workflow. It is designed around EBSD patterns: vendor-file access, lazy handling of large pattern arrays, detector and projection geometry, background correction, pattern processing, simulation-related operations, and indexing interoperability. Its data model uses the HyperSpy ecosystem and orix crystallographic objects.\n\n**Grains and texture.** It should not be selected as the principal grain-reconstruction engine. Once indexing has produced orientation, phase, quality, and coordinate arrays, downstream segmentation and grain statistics generally require another package or custom code. Likewise, kikuchipy is not a comprehensive ODF/texture-analysis system and does not offer an MTEX-equivalent grain/boundary object model.\n\n**KAM/GND.** Neither is a core turnkey kikuchipy capability. KAM can be calculated from its output arrays with orix/NumPy-style operations, but masking, symmetry reduction, neighborhood definition, and exclusion of high-angle boundaries remain the pipeline developer\u2019s responsibility. GND inversion is substantially further outside its main scope.\n\n**Maturity.** kikuchipy is a well-developed, domain-specific open-source project with a software paper titled *kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns*. Its maturity is strongest for pattern-space operations, not for post-indexing microstructure analysis.\n\n### 2. orix\n\n**Role.** orix is the most important Python-native crystallographic foundation. It supplies rotations and orientations, crystal and specimen symmetry, fundamental-zone reduction, symmetry-aware misorientation, orientation colours, inverse-pole-figure colour keys, pole-figure-style projections, and orientation clustering. Its peer-reviewed methodological reference is Johnstone et al., *Density-based clustering of crystal (mis)orientations and the orix Python library*, *Journal of Applied Crystallography* 53 (2020), 1293\u20131298, DOI 10.1107/S1600576720011103.\n\n**Grain reconstruction.** orix contains the primitives needed to compare neighboring orientations and cluster orientation data, but it is not yet an MTEX-like turnkey EBSD topology package. A production grain segmenter still needs scan-grid connectivity, phase masks, confidence masks, flood-fill or graph labeling, small-region handling, and grain-property aggregation. Those steps can be implemented with NumPy, SciPy, scikit-image, xarray, and networkx, but then the laboratory owns validation and maintenance.\n\n**Texture.** orix is strong for IPF colouring, stereographic projections, symmetry handling, and discrete orientation distributions. It is suitable for PF/IPF visualization and custom orientation-density estimation. Nevertheless, MTEX remains substantially richer and more established for quantitative ODF estimation, kernel-selection conventions, harmonic representations, fibres/components, uncertainty-aware texture workflows, and publication-ready texture analysis. An ODF reconstructed from discrete EBSD orientations is also parameter-dependent; published work notes sensitivity to ODF computation settings in MTEX itself, so Python/MTEX agreement requires matching kernels, bandwidths, symmetry, weights, and normalization. A modern EBSD texture study used pole figures to characterize weak texture and illustrates the need to retain a statistically representative orientation population. (pqac-00000006)\n\n**KAM/GND.** orix is a good basis for KAM because it provides symmetry-aware misorientation operations. It does not, by itself, turn KAM into a one-call validated EBSD workflow. GND requires spatial orientation gradients, reference-frame consistency, Burgers vectors/slip-system assumptions, Nye-tensor construction, and an underdetermined inversion; this remains custom research code rather than an MTEX-equivalent orix feature.\n\n### 3. DefDAP\n\n**Role.** DefDAP\u2014Deformation Data Analysis in Python\u2014is a Python package aimed at combining EBSD with deformation measurements such as high-resolution digital image correlation. Its citable software release is Atkinson et al., *DefDAP: Deformation Data Analysis in Python* (2020), DOI 10.5281/zenodo.3688097.\n\n**Grains.** DefDAP is useful for 2D EBSD maps, grain finding and grain-level analysis, crystallographic slip/twinning interpretation, boundary overlays, and linking deformation fields to grains. For a conventional single-phase 2D Al EBSD map it may be the quickest lightweight route from an orientation map to grains and grain-wise quantities.\n\n**Texture and misorientation metrics.** Its emphasis is local deformation and grain-scale mechanics, not comprehensive quantitative texture. It is less complete than MTEX for ODFs, texture components, and generalized boundary/reconstruction operations. KAM-like local misorientation calculations can be added or may be available through map-level operations depending on release, but GND-density inversion should not be assumed to be a complete, standardized capability without checking the exact version and validating the implementation.\n\n**Maturity.** DefDAP has been used in peer-reviewed deformation studies, but its community, documentation breadth, and EBSD feature coverage are smaller than MTEX or DREAM.3D. It is best regarded as a focused 2D deformation-analysis package rather than a universal EBSD backend.\n\n### 4. DREAM.3D and DREAM3D-NX/`simplnx`\n\n**Role and architecture.** Legacy DREAM.3D is an established C++/Qt microstructure-analysis platform. DREAM3D-NX is its current-generation successor; `simplnx` provides Python bindings and pipeline construction. Thus, it is **Python-accessible and scriptable**, although not pure Python internally. This distinction is normally beneficial for large EBSD maps because segmentation and feature-statistics filters execute in compiled code.\n\n**Grain reconstruction.** This is the strongest open-source alternative for automated grain segmentation. DREAM.3D uses pipeline filters for data import, masking, cleanup, symmetry-aware misorientation segmentation, feature identification, neighbor lists, grain morphology, crystallographic statistics, boundary meshes, and, where applicable, 3D reconstruction. Published work describes DREAM.3D as a unified data structure aggregating community-developed algorithms, with reconstruction pipelines commonly containing more than a dozen parameterized steps. (pqac-00000011)\n\nIts `Segment Features (Misorientation)` workflow has been used to compare 2D and 3D segmentation across several misorientation thresholds. (pqac-00000007) Published AM/TriBeam workflows reconstructed serial EBSD slices, segmented grains using explicit misorientation tolerances, removed features below specified voxel counts, filled gaps, and automated parameter sweeps. (pqac-00000003, pqac-00000010) Cloud integration further demonstrated parallel parameter studies, provenance tracking, and grain-size comparisons between 2D and 3D AM datasets. (pqac-00000001, pqac-00000002)\n\n**Texture.** DREAM.3D calculates crystallographic feature statistics and supports orientation-distribution data used in synthetic microstructure generation, but it is not as strong as MTEX for interactive quantitative texture analysis. PF/IPF maps and basic orientation statistics are feasible; sophisticated ODF reconstruction and texture-component analysis are better handled in orix plus custom numerical code\u2014or validated against MTEX. Published workflows frequently use DREAM.3D for microstructure generation/reconstruction and MTEX for detailed ODF/PF analysis, reflecting this division of strengths. (pqac-00000000)\n\n**KAM/GND.** DREAM3D-NX should not be selected primarily for KAM or GND. Depending on installed plugins and version, local orientation filters may cover some neighborhood measures, but these capabilities are less central and less transparent than its feature segmentation/statistics pipeline. A NumPy/orix KAM stage is easier to audit. GND remains a specialist calculation.\n\n**Cautions.** Segmentation is not parameter-free. The same threshold can produce different 2D and 3D connectivity, especially for nonconvex grains or paths connected outside the selected plane. (pqac-00000005, pqac-00000007, pqac-00000008) Minimum feature size and cleanup can also shift the measured grain-size distribution. Consequently, pipeline JSON, threshold, neighborhood convention, phase mask, and deleted/fill operations should be retained as provenance.\n\n### 5. pyxem\n\npyxem is an open-source Python ecosystem for multidimensional diffraction microscopy, built around HyperSpy and commonly used for scanning/precession electron diffraction and 4D-STEM. It is valuable for virtual diffraction imaging, diffraction-vector analysis, orientation mapping, and related TEM workflows.\n\nIt is **not a direct downstream package for conventional SEM-EBSD orientation maps**. Grain segmentation from vendor-style EBSD grids, quantitative EBSD texture analysis, KAM, and GND are not its central abstractions. Its shared ecosystem with kikuchipy/orix can be useful if the project later combines EBSD with scanning electron diffraction, but pyxem should not be the primary AlSi10Mg grain-analysis backend.\n\n### 6. PyMicro\n\nPyMicro provides Python classes and algorithms for crystallography, microstructure representation, grains, orientations, meshes, and links to simulation and tomography. It can import or represent EBSD-like microstructures and is useful for custom data integration, synthetic microstructures, and computational-mechanics coupling.\n\nFor this task, its limitations are that conventional EBSD segmentation, comprehensive ODF analysis, KAM, and GND are not offered with the same turnkey breadth or community validation as MTEX. It is better viewed as an extensible microstructure framework than as a drop-in EBSD analysis application.\n\n### 7. OpenXY and other research code\n\nOpenXY is associated mainly with high-angular-resolution EBSD cross-correlation and strain/rotation analysis. It addresses a different problem from ordinary Hough-indexed orientation-map characterization and is not a complete grain/texture package. Some Python ports and research repositories calculate lattice curvature or GND density, but their conventions, slip-system bases, regularization, and maintenance vary.\n\nOther small packages carrying names such as `pyebsd`, custom scikit-image segmenters, or laboratory scripts can calculate IPF colours, KAM, or threshold-connected grains. They can be useful building blocks, but package names are sometimes ambiguous, releases may be unmaintained, and peer-reviewed software references are often absent. They should not be treated as equivalent to MTEX merely because they produce an IPF or KAM image.\n\n## Recommended pipeline for AM AlSi10Mg\n\n### Data contract after indexing\n\nExport or retain the following arrays and metadata:\n\n- quaternion or rotation matrix per scan point\u2014preferably not only Euler angles;\n- phase ID, indexed/unindexed mask, confidence or match score, and pattern quality;\n- physical `x`, `y` coordinates and step size;\n- scan shape and rectangular/hexagonal topology;\n- crystal symmetry and lattice parameters for fcc Al;\n- detector/sample/reference-frame conventions;\n- indexing dictionary or phase-library version.\n\nAlSi10Mg is commonly dominated in EBSD by fcc \u03b1-Al, while the fine Si network may be incompletely indexed at ordinary step sizes. Do not let unindexed Si-rich pixels automatically split Al grains: establish an explicit phase/mask and cleanup policy and report it.\n\n### Grain stage\n\nUse **DREAM3D-NX/`simplnx`** for the production pipeline:\n\n1. construct the image geometry and load quaternion, phase, mask, and quality arrays;\n2. remove only demonstrably unreliable points;\n3. segment same-phase connected pixels with a documented disorientation threshold;\n4. calculate feature centroids, equivalent diameters, areas, aspect ratios, mean orientations, neighbors, and boundary misorientations;\n5. retain both the raw feature labels and any post-cleanup labels;\n6. sweep plausible thresholds rather than fixing one unexamined value.\n\nFor AM AlSi10Mg, evaluate at least a low-angle-sensitive threshold and a conventional grain threshold\u2014for example, a controlled sweep around 2\u20135\u00b0\u2014because cellular/subgrain orientation gradients may be physically meaningful. This is not a recommendation to choose a universal value; published work demonstrates that changing tolerance can radically alter segmentation and that 2D and 3D results are not automatically equivalent. (pqac-00000003, pqac-00000005, pqac-00000007)\n\nIf only modest 2D maps are involved and direct Python object manipulation is more important than DREAM3D-NX pipelines, **DefDAP** is a reasonable alternative. A fully custom orix/scikit-image segmenter is defensible only if regression-tested against DREAM3D-NX or MTEX.\n\n### Texture stage\n\nUse **orix** for orientation conversion, symmetry reduction, IPF maps, pole figures, and discrete orientation statistics. For ODF estimation:\n\n- weight points or grains deliberately; pixel-weighted and one-orientation-per-grain ODFs answer different questions;\n- state the kernel and half-width/bandwidth;\n- enforce cubic crystal symmetry and the intended specimen symmetry;\n- normalize PF/ODF intensities consistently, usually in multiples of random distribution;\n- validate against a known texture or a small MTEX reference dataset.\n\nThis distinction matters in AM material: pixel weighting emphasizes large columnar grains, whereas grain weighting describes the population of grains. A recent study highlights the importance of preserving the full orientation distribution when reducing EBSD orientation datasets and uses PF maps to assess texture strength. (pqac-00000006)\n\n### KAM\n\nKAM is straightforward to implement but easy to define inconsistently. For each indexed point, compute the mean symmetry-reduced disorientation to valid neighbors, while specifying:\n\n- first-, second-, or higher-order neighborhood;\n- whether diagonals are included;\n- maximum included disorientation, often chosen to exclude grain boundaries;\n- treatment of phase boundaries, map edges, and unindexed pixels;\n- degrees versus radians;\n- whether the map has been denoised before calculation.\n\nUse orix for disorientation and NumPy/SciPy for neighborhood operations. Store the number of valid neighbors along with KAM; otherwise values near pores, Si particles, and map boundaries are misleading. KAM should not be interpreted directly as dislocation density.\n\n### GND density\n\nGND is the least turnkey part of a MATLAB-free pipeline. A defensible calculation needs orientation gradients in physical units, lattice curvature, an explicit Nye-tensor convention, Burgers-vector/slip-system definitions, and a solution to an underdetermined inversion\u2014commonly L1 or L2 minimization. Report whether the result is total scalar GND density, a lower bound, or slip-system-resolved density.\n\nFor conventional EBSD, noise and angular resolution strongly affect spatial derivatives. The result also scales with step size and preprocessing. Therefore:\n\n1. preserve orientations at full precision;\n2. estimate angular noise on an undeformed/reference region;\n3. document derivative stencil and smoothing;\n4. perform step-size or downsampling sensitivity tests;\n5. benchmark synthetic rotation-gradient fields with known curvature;\n6. compare a representative map against an established MTEX or HR-EBSD implementation.\n\nNo named Python package in this comparison offers a broadly validated, turnkey GND workflow matching MTEX\u2019s integration. Treat GND as an independently versioned scientific module, not as a plotting option attached to the grain segmenter.\n\n## Practical ranking\n\n- **Best production grain segmentation and statistics:** DREAM3D-NX/`simplnx`.\n- **Best Python crystallographic foundation and PF/IPF layer:** orix.\n- **Best upstream pattern/indexing integration:** kikuchipy + PyEBSDIndex.\n- **Best lightweight 2D grain/deformation package:** DefDAP.\n- **Best for TEM diffraction-microscopy integration, not SEM-EBSD post-processing:** pyxem.\n- **Best custom microstructure/simulation framework:** PyMicro.\n- **No fully satisfactory turnkey Python winner for GND:** implement and validate separately.\n\nAccordingly, the recommended deployable stack is **kikuchipy/PyEBSDIndex \u2192 DREAM3D-NX/`simplnx` \u2192 orix-based texture and KAM \u2192 validated custom GND module**. This removes MATLAB from routine production while retaining MTEX only as an optional one-time validation reference. DREAM.3D\u2019s documented use on complex additively manufactured 3D EBSD datasets, including automated parameter studies and explicit misorientation segmentation, makes this architecture substantially lower-risk than building the entire downstream stage directly from NumPy/orix primitives. (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000010, pqac-00000011)", + "answer_reasoning": null, + "has_successful_answer": true, + "context": "", + "contexts": [ + { + "id": "pqac-0bc07815", + "context": "This excerpt describes MTEX as a MATLAB toolbox capable of analyzing crystallographic texture through inverse pole figure (IPF) maps and orientation distribution function (ODF) pole figures. The tool can extract and analyze grain size, shape, and orientation properties of microstructures by fitting grains with ellipses. MTEX provides standard functions for grain size analysis (calculating equivalent circle diameters fitted to lognormal distributions), aspect ratio calculations from ellipse axes, and a characteristic shape function that sums over grain boundary segment orientations. The excerpt documents use of MTEX for texture validation in synthetic microstructure analysis. However, the excerpt does not provide information about Python interfaces, open-source status, KAM/GND calculations, grain segmentation/reconstruction capabilities, or the maturity/publication record of MTEX or other listed tools (kikuchipy, orix, DefDAP, DREAM.3D, pyxem, PyMicro).", + "question": "Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.", + "text": { + "embedding": null, + "text": "", + "name": "folwarczny2026accurategrainboundary pages 5-9", + "media": [], + "doc": { + "embedding": null, + "docname": "folwarczny2026accurategrainboundary", + "dockey": "59d8aa55a8c3757a", + "citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "folwarczny2026accurategrainboundary", + "bibtex": "@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n", + "authors": [ + "Martin Folwarczny", + "Ao Li", + "Rushvi Shah", + "Aaron Chote", + "Alexandra C. Austin", + "Yimin Zhu", + "Gregory S. Rohrer", + "Michael A. Jackson", + "Souhardh Kotakadi", + "Katharina Marquardt" + ], + "publication_date": "2026-02-01T00:00:00Z", + "year": 2026, + "volume": "280", + "issue": null, + "issn": "0304-3991", + "pages": "114262", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "title": "Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2025.114262", + "doi_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "doc_id": "59d8aa55a8c3757a", + "file_location": null, + "license": null, + "pdf_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal." + }, + "chunk_id": "8780a7be-14a2-4832-9f6b-1beb6ba2ee0a" + }, + "score": 4, + "snippets": [ + "The first step in the analysis of the synthetic orientation maps was performed using the MTEX toolbox for MATLAB [33].", + "Standard functions provided by MTEX were used to plot inverse pole figure (IPF) maps and orientation distribution function (ODF) pole figures to analyse the crystallographic texture of the simulated microstructure.", + "The grain size, shape and orientation properties of the microstructure were also analysed using MTEX, fitting grains with ellipses.", + "The characteristic shape function is provided by MTEX and sums over grain boundary segment orientations.", + "A MATLAB script was written to convert these files into orientation maps in the .ang format.", + "Grain diameters were calculated and fitted with a lognormal distribution." + ], + "used_images": false + }, + { + "id": "pqac-6926a3e1", + "context": "This excerpt describes a stereological method for obtaining grain boundary plane distributions (GBPD) from 2D EBSD orientation maps applied to textured microstructures. The work references code from Carnegie Mellon University's Rohrer group for stereological analysis and uses DREAM.3D for synthetic microstructure generation. However, the excerpt does not provide detailed information about the capabilities, open-source status, Python interfaces, maturity, publications, or specific computational features (such as grain segmentation, texture analysis, KAM, or GND calculations) of kikuchipy, orix, DefDAP, DREAM.3D, pyxem, MTEX, PyMicro, or other Python EBSD tools. The document focuses on presenting a novel stereological approach rather than surveying existing EBSD analysis software.", + "question": "Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.", + "text": { + "embedding": null, + "text": "", + "name": "folwarczny2026accurategrainboundary pages 1-3", + "media": [], + "doc": { + "embedding": null, + "docname": "folwarczny2026accurategrainboundary", + "dockey": "59d8aa55a8c3757a", + "citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "folwarczny2026accurategrainboundary", + "bibtex": "@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n", + "authors": [ + "Martin Folwarczny", + "Ao Li", + "Rushvi Shah", + "Aaron Chote", + "Alexandra C. Austin", + "Yimin Zhu", + "Gregory S. Rohrer", + "Michael A. Jackson", + "Souhardh Kotakadi", + "Katharina Marquardt" + ], + "publication_date": "2026-02-01T00:00:00Z", + "year": 2026, + "volume": "280", + "issue": null, + "issn": "0304-3991", + "pages": "114262", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "title": "Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2025.114262", + "doi_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "doc_id": "59d8aa55a8c3757a", + "file_location": null, + "license": null, + "pdf_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal." + }, + "chunk_id": "4287fc76-69ba-485a-b51c-80b963650c1d" + }, + "score": 2, + "snippets": [ + "Stereology, applied to 2D EBSD orientation maps, is currently the fastest method of obtaining GBPDs", + "code written at Carnegie Mellon University in the Rohrer group", + "We combined microstructure simulation of a textured material using DREAM.3D" + ], + "used_images": false + }, + { + "id": "pqac-c92f0d5c", + "context": "The excerpt mentions MTEX as a toolbox available for MATLAB that is adept at handling and interpreting EBSD data, specifically for extracting Euler angles from EBSD scans. The paper describes using MTEX for initial EBSD dataset preprocessing, followed by clustering performed in Python. The excerpt also references the orientation distribution function (ODF) as a method that can be employed for analysis. However, the excerpt does not provide detailed information about the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction capabilities, texture analysis details (ODF/pole figures/IPF), KAM, or GND calculations of MTEX or other EBSD tools like kikuchipy, orix, DefDAP, DREAM.3D, pyxem, or PyMicro.", + "question": "Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.", + "text": { + "embedding": null, + "text": "", + "name": "wanni2024machinelearningenhanced pages 6-7", + "media": [], + "doc": { + "embedding": null, + "docname": "wanni2024machinelearningenhanced", + "dockey": "bf3e624ed86e8966", + "citation": "J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "wanni2024machinelearningenhanced", + "bibtex": "@article{wanni2024machinelearningenhanced,\n author = \"Wanni, J. and Bronkhorst, C. A. and Thoma, D. J.\",\n title = \"Machine learning enhanced analysis of EBSD data for texture representation\",\n year = \"2024\",\n journal = \"npj Computational Materials\",\n volume = \"10\",\n pages = \"1-11\",\n month = \"Jun\",\n doi = \"10.1038/s41524-024-01324-4\",\n url = \"https://doi.org/10.1038/s41524-024-01324-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n", + "authors": [ + "J. Wanni", + "C. A. Bronkhorst", + "D. J. Thoma" + ], + "publication_date": "2024-06-26T00:00:00Z", + "year": 2024, + "volume": "10", + "issue": "1", + "issn": "2057-3960", + "pages": "1-11", + "journal": "npj Computational Materials", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41524-024-01324-4", + "title": "Machine learning enhanced analysis of EBSD data for texture representation", + "citation_count": 14, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41524-024-01324-4", + "doi_url": "https://doi.org/10.1038/s41524-024-01324-4", + "doc_id": "bf3e624ed86e8966", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41524-024-01324-4.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4. This article has 14 citations and is from a peer-reviewed journal." + }, + "chunk_id": "3f64266c-1c24-4c26-92ca-e7123133e273" + }, + "score": 3, + "snippets": [ + "The EBSD raw datasets were pre-processed using the MTEX toolbox available for MATLAB and the clustering was performed in the python environment.", + "MTEX is adept at handling and interpreting the complex data contained in EBSD scans, making it an ideal choice for the initial processing phase.", + "This conversion to .npy files is a critical step as it opens up the data to the extensive ecosystem of Python libraries and tools, particularly those tailored for data analysis and machine learning.", + "The orientation distribution function (ODF) also can be employed to determine the optimal number of data points." + ], + "used_images": false + }, + { + "id": "pqac-a9c3bc56", + "context": "This excerpt describes the use of DREAM.3D software for synthetic microstructure generation with textured materials. The paper demonstrates DREAM.3D's capabilities for: (1) grain shape simulation using ellipsoid axis ratios, (2) grain size distribution specification via lognormal distributions with mean and standard deviation parameters, (3) crystallographic texture definition through Euler angles in Bunge convention with weight assignments, and (4) crystal symmetry specification for different phases (hexagonal Zircaloy-4 and tetragonal zirconia). The excerpt also mentions MTEX's use for grain size metric calculation (equivalent circle diameter). However, the excerpt provides limited detail on open-source status, Python interfaces, maturity level, publications, grain segmentation/reconstruction capabilities, ODF/pole figure generation, KAM, GND calculations, or comparative information about other Python EBSD tools (kikuchipy, orix, DefDAP, pyxem, PyMicro). The document focuses on methodology for validating stereological GBPD (grain boundary plane distribution) analysis rather than comprehensive tool characterization.", + "question": "Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.", + "text": { + "embedding": null, + "text": "", + "name": "folwarczny2026accurategrainboundary pages 3-5", + "media": [], + "doc": { + "embedding": null, + "docname": "folwarczny2026accurategrainboundary", + "dockey": "59d8aa55a8c3757a", + "citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "folwarczny2026accurategrainboundary", + "bibtex": "@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n", + "authors": [ + "Martin Folwarczny", + "Ao Li", + "Rushvi Shah", + "Aaron Chote", + "Alexandra C. Austin", + "Yimin Zhu", + "Gregory S. Rohrer", + "Michael A. Jackson", + "Souhardh Kotakadi", + "Katharina Marquardt" + ], + "publication_date": "2026-02-01T00:00:00Z", + "year": 2026, + "volume": "280", + "issue": null, + "issn": "0304-3991", + "pages": "114262", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "title": "Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2025.114262", + "doi_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "doc_id": "59d8aa55a8c3757a", + "file_location": null, + "license": null, + "pdf_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal." + }, + "chunk_id": "1b1f7b0d-a04c-4ba3-abc9-7f7db7412bcb" + }, + "score": 3, + "snippets": [ + "we generated the textured synthetic mono-phased microstructures using DREAM.3D", + "The grain size was specified using a lognormal distribution", + "The crystallographic texture was set by specifying orientations via Euler angles in the Bunge convention", + "The grain size metric used in MTEX was equivalent circle diameter (ECD)" + ], + "used_images": false + }, + { + "id": "pqac-de7dc608", + "context": "The excerpt provides limited information about Python EBSD tools. It contains two relevant references: MTEX is described as a 'Free and Open Source Software Toolbox' focused on texture analysis, published in 2010. PyMicro is referenced as a 2021 resource by Proudhon. The excerpt also mentions various diffraction contrast tomography and electron backscatter diffraction applications in materials characterization, but does not provide detailed capabilities, functionality comparisons, or specific information about kikuchipy, orix, DefDAP, DREAM.3D, pyxem, or other Python EBSD tools. The references cited relate more to grain reconstruction, diffraction contrast tomography, and general materials characterization rather than comprehensive documentation of Python EBSD software capabilities.", + "question": "Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.", + "text": { + "embedding": null, + "text": "", + "name": "ball2023registrationbetweendct pages 14-15", + "media": [], + "doc": { + "embedding": null, + "docname": "ball2023registrationbetweendct", + "dockey": "0a036b5088d09190", + "citation": "J. Ball, J. Oddershede, C. Davis, C. Slater, M. Said, H. Vashishtha, \u0160. Michalik, David M. Collins School of Metallurgy, Materials, U. Birmingham, United Kingdom., Diamond Light Source Ltd., Xnovo Technology ApS, Denmark, Wmg, and U. Warwick. Registration between dct and ebsd datasets for multiphase microstructures. Materials Characterization, 204:113228, Oct 2023. URL: https://doi.org/10.1016/j.matchar.2023.113228, doi:10.1016/j.matchar.2023.113228.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "ball2023registrationbetweendct", + "bibtex": "@article{ball2023registrationbetweendct,\n author = \"Ball, J. and Oddershede, J. and Davis, C. and Slater, C. and Said, M. and Vashishtha, H. and Michalik, \u0160. and of Metallurgy, David M. Collins School and Materials and Birmingham, U. and Kingdom., United and Ltd., Diamond Light Source and ApS, Xnovo Technology and Denmark and Wmg and Warwick, U.\",\n title = \"Registration between DCT and EBSD datasets for multiphase microstructures\",\n year = \"2023\",\n journal = \"Materials Characterization\",\n volume = \"204\",\n pages = \"113228\",\n month = \"Oct\",\n doi = \"10.1016/j.matchar.2023.113228\",\n url = \"https://doi.org/10.1016/j.matchar.2023.113228\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n", + "authors": [ + "J. Ball", + "J. Oddershede", + "C. Davis", + "C. Slater", + "M. Said", + "H. Vashishtha", + "\u0160. Michalik", + "David M. Collins School of Metallurgy", + "Materials", + "U. Birmingham", + "United Kingdom.", + "Diamond Light Source Ltd.", + "Xnovo Technology ApS", + "Denmark", + "Wmg", + "U. Warwick" + ], + "publication_date": "2023-10-01T00:00:00Z", + "year": 2023, + "volume": "204", + "issue": null, + "issn": "1044-5803", + "pages": "113228", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2023.113228", + "title": "Registration between DCT and EBSD datasets for multiphase microstructures", + "citation_count": 8, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2023.113228", + "doi_url": "https://doi.org/10.1016/j.matchar.2023.113228", + "doc_id": "0a036b5088d09190", + "file_location": null, + "license": null, + "pdf_url": "https://ars.els-cdn.com/content/image/1-s2.0-S1044580323005879-ga1_lrg.jpg", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Ball, J. Oddershede, C. Davis, C. Slater, M. Said, H. Vashishtha, \u0160. Michalik, David M. Collins School of Metallurgy, Materials, U. Birmingham, United Kingdom., Diamond Light Source Ltd., Xnovo Technology ApS, Denmark, Wmg, and U. Warwick. Registration between dct and ebsd datasets for multiphase microstructures. Materials Characterization, 204:113228, Oct 2023. URL: https://doi.org/10.1016/j.matchar.2023.113228, doi:10.1016/j.matchar.2023.113228. This article has 8 citations and is from a peer-reviewed journal." + }, + "chunk_id": "451fa479-150b-47db-ad7e-9db5e58bb3f2" + }, + "score": 3, + "snippets": [ + "F. Bachmann, H. Bale, N. Gueninchault, C. Holzner, E. M. Lauridsen, Texture Analysis with MTEX \u2013 Free and Open Source Software Toolbox, Solid State Phenomena 160 (2010) 63\u201368.", + "H. Proudhon, pymicro, 2021." + ], + "used_images": false + }, + { + "id": "pqac-81ba8eaa", + "context": "The excerpt contains bibliography references to three relevant software tools mentioned in the question. MTEX is described as free and open source software for texture analysis (Bachmann et al. 2010). DREAM.3D is identified as a digital representation environment for 3D microstructure analysis (Groeber & Jackson 2014). ORIX is referenced as a Python library for density-based clustering of crystal (mis)orientations (Johnstone et al. 2020). However, the excerpt provides only citation information without detailed descriptions of these tools' specific capabilities, maturity levels, Python interfaces, grain segmentation/reconstruction methods, ODF/pole figure generation, KAM or GND calculation features, or their limitations. The reference list does not contain information about kikuchipy, DefDAP, DREAM3D-NX/simplnx, pyxem, or PyMicro.", + "question": "Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2026pytribeamopensourcesoftware pages 16-17", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2026pytribeamopensourcesoftware", + "dockey": "4deb79e3fdff92cc", + "citation": "Andrew T. Polonsky, James D. Lamb, Chad B. Hovey, Haydn Schroader, McLean P. Echlin, and Tresa M. Pollock. Pytribeam: open-source software for enhanced 3d data collection in tribeam microscopes. Integrating Materials and Manufacturing Innovation, 15:1-17, Feb 2026. URL: https://doi.org/10.1007/s40192-026-00449-2, doi:10.1007/s40192-026-00449-2.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2026pytribeamopensourcesoftware", + "bibtex": "@article{polonsky2026pytribeamopensourcesoftware,\n author = \"Polonsky, Andrew T. and Lamb, James D. and Hovey, Chad B. and Schroader, Haydn and Echlin, McLean P. and Pollock, Tresa M.\",\n title = \"pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes\",\n year = \"2026\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"15\",\n pages = \"1-17\",\n month = \"Feb\",\n doi = \"10.1007/s40192-026-00449-2\",\n url = \"https://doi.org/10.1007/s40192-026-00449-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "James D. Lamb", + "Chad B. Hovey", + "Haydn Schroader", + "McLean P. Echlin", + "Tresa M. Pollock" + ], + "publication_date": "2026-02-27T00:00:00Z", + "year": 2026, + "volume": "15", + "issue": "1", + "issn": "2193-9764", + "pages": "1-17", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-026-00449-2", + "title": "pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-026-00449-2", + "doi_url": "https://doi.org/10.1007/s40192-026-00449-2", + "doc_id": "4deb79e3fdff92cc", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-026-00449-2.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, James D. Lamb, Chad B. Hovey, Haydn Schroader, McLean P. Echlin, and Tresa M. Pollock. Pytribeam: open-source software for enhanced 3d data collection in tribeam microscopes. Integrating Materials and Manufacturing Innovation, 15:1-17, Feb 2026. URL: https://doi.org/10.1007/s40192-026-00449-2, doi:10.1007/s40192-026-00449-2. This article has 0 citations and is from a peer-reviewed journal." + }, + "chunk_id": "87a15f80-531e-420a-b10d-144810a80182" + }, + "score": 3, + "snippets": [ + "Bachmann F, Hielscher R, Schaeben H (2010) Texture analysis with mtex \u2013 free and open source software toolbox. Solid State Phenom 160:63\u201368", + "Groeber MA, Jackson MA (2014) Dream. 3d: A digital representation environment for the analysis of microstructure in 3D. Integrating Materials and Manufacturing Innovation 3(1):56\u201372", + "Johnstone DN, Martineau BH, Crout P, Midgley PA, Eggeman AS (2020) Density-based clustering of crystal (mis)orientations and the orix python library. J Appl Crystallogr 53(5):1293\u20131298" + ], + "used_images": false + }, + { + "id": "pqac-8d9d9db3", + "context": "This excerpt discusses the use of KAM (Kernel Average Misorientation) and pattern quality as crystallographic features derived from EBSD data for microstructural analysis. The paper mentions that KAM and pattern quality have been used as input parameters to machine learning algorithms for steel microstructure classification, and demonstrates that orientation-derived parameters such as KAM achieved ~98% pixel-wise accuracy in their CNN-based phase segmentation model. However, the excerpt does not provide information about specific Python-based EBSD analysis tools (kikuchipy, orix, DefDAP, DREAM.3D, pyxem, MTEX, PyMicro) or their capabilities, open-source status, grain segmentation/reconstruction features, texture analysis capabilities, or GND calculations.", + "question": "Across the retrieved literature, what evidence describes the capabilities, open-source status, Python interfaces, maturity, publications, grain segmentation/reconstruction, texture analysis (ODF/pole figures/IPF), KAM, and GND calculations of kikuchipy, orix, DefDAP, DREAM.3D or DREAM3D-NX/simplnx, pyxem, MTEX, PyMicro, and other Python EBSD tools? Extract exact statements and bibliographic details, including limitations.", + "text": { + "embedding": null, + "text": "", + "name": "ostormujof2022deeplearningfor pages 1-3", + "media": [], + "doc": { + "embedding": null, + "docname": "ostormujof2022deeplearningfor", + "dockey": "ddc178136b11c42d", + "citation": "T Martinez Ostormujof, Rrp Purushottam Raj Purohit, S Breumier, Nathalie Gey, M Salib, and L Germain. Deep learning for automated phase segmentation in ebsd maps. a case study in dual phase steel microstructures. Materials Characterization, Nov 2022. URL: https://doi.org/10.48550/arxiv.2112.03072, doi:10.48550/arxiv.2112.03072.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "ostormujof2022deeplearningfor", + "bibtex": "@article{ostormujof2022deeplearningfor,\n author = \"Ostormujof, T Martinez and Purohit, Rrp Purushottam Raj and Breumier, S and Gey, Nathalie and Salib, M and Germain, L\",\n title = \"Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures\",\n year = \"2022\",\n journal = \"Materials Characterization\",\n month = \"Nov\",\n doi = \"10.48550/arxiv.2112.03072\",\n url = \"https://doi.org/10.48550/arxiv.2112.03072\",\n publisher = \"arXiv\"\n}\n", + "authors": [ + "T Martinez Ostormujof", + "Rrp Purushottam Raj Purohit", + "S Breumier", + "Nathalie Gey", + "M Salib", + "L Germain" + ], + "publication_date": "2021-11-26T00:00:00Z", + "year": 2022, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Materials Characterization", + "publisher": "arXiv", + "url": "https://arxiv.org/abs/2112.03072", + "title": "Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures", + "citation_count": 65, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.48550/arxiv.2112.03072", + "doi_url": "https://doi.org/10.48550/arxiv.2112.03072", + "doc_id": "ddc178136b11c42d", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1044580321007609", + "other": { + "client_source": [ + "paper_search", + "datacite", + "semantic_scholar", + "datacite", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "T Martinez Ostormujof, Rrp Purushottam Raj Purohit, S Breumier, Nathalie Gey, M Salib, and L Germain. Deep learning for automated phase segmentation in ebsd maps. a case study in dual phase steel microstructures. Materials Characterization, Nov 2022. URL: https://doi.org/10.48550/arxiv.2112.03072, doi:10.48550/arxiv.2112.03072. This article has 65 citations and is from a peer-reviewed journal." + }, + "chunk_id": "522c5142-1f9e-4462-8b4e-b2e40ddb1497" + }, + "score": 2, + "snippets": [ + "Kernel Average Misorientation (KAM) or pattern quality", + "Crystallographic features obtained by EBSD as variant pair density, Kernel Average Misorientation (KAM) and pattern quality have also been fed to ML algorithms to classify steel's microstructures", + "parameters derived from orientation data as pattern quality and KAM" + ], + "used_images": false + }, + { + "id": "pqac-00000000", + "context": "The excerpt describes using MTEX and DREAM.3D for EBSD analysis. MTEX capabilities include: plotting inverse pole figure (IPF) maps and orientation distribution function (ODF) pole figures for texture analysis, grain size distribution analysis using equivalent circle radius function with lognormal fitting, grain shape analysis via ellipse fitting with aspect ratio and characteristic shape function calculations, and grain boundary segment orientation analysis. DREAM.3D is used for simulating 3D microstructures and exporting 2D planar sections in multiple orthogonal directions (XY, XZ, YZ) with proper reference frame rotations to match experimental EBSD conventions. The workflow converts DREAM.3D output to standard .ang format used in TSL OIM Analysis. No information is provided regarding KAM usage, orix, DefDAP, kikuchipy, pyxem, or PyMicro tools.", + "question": "Extract evidence relevant to comparing EBSD post-processing tools: (1) MTEX grain reconstruction and texture/ODF capabilities; (2) DREAM.3D grain segmentation and 2D/3D EBSD processing; (3) use of KAM or grain boundaries in EBSD workflows; (4) peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, PyMicro. Include exact capabilities and limitations where stated.", + "text": { + "embedding": null, + "text": "", + "name": "folwarczny2026accurategrainboundary pages 5-9", + "media": [], + "doc": { + "embedding": null, + "docname": "folwarczny2026accurategrainboundary", + "dockey": "59d8aa55a8c3757a", + "citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "folwarczny2026accurategrainboundary", + "bibtex": "@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n", + "authors": [ + "Martin Folwarczny", + "Ao Li", + "Rushvi Shah", + "Aaron Chote", + "Alexandra C. Austin", + "Yimin Zhu", + "Gregory S. Rohrer", + "Michael A. Jackson", + "Souhardh Kotakadi", + "Katharina Marquardt" + ], + "publication_date": "2026-02-01T00:00:00Z", + "year": 2026, + "volume": "280", + "issue": null, + "issn": "0304-3991", + "pages": "114262", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "title": "Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2025.114262", + "doi_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "doc_id": "59d8aa55a8c3757a", + "file_location": null, + "license": null, + "pdf_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal." + }, + "chunk_id": "8780a7be-14a2-4832-9f6b-1beb6ba2ee0a" + }, + "score": 5, + "snippets": [ + "The first step in the analysis of the synthetic orientation maps was performed using the MTEX toolbox for MATLAB [33]. To verify the input parameters for the DREAM.3D simulations, the crystallographic texture, grain size distribution, grain shape and orientations were extracted and compared to the DREAM.3D input.", + "Standard functions provided by MTEX were used to plot inverse pole figure (IPF) maps and orientation distribution function (ODF) pole figures to analyse the crystallographic texture of the simulated microstructure.", + "The grain size, shape and orientation properties of the microstructure were also analysed using MTEX, fitting grains with ellipses. Grain size was defined using the equivalent circle radius function in MTEX.", + "The grain shape was analysed in two ways: aspect ratio and characteristic shape function. Results are plotted in Figure 5 for the hexagonal microstructure and Figure 6 for the tetragonal. The aspect ratios were calculated by comparing the lengths of the ellipse long and short axes. The characteristic shape function is provided by MTEX and sums over grain boundary segment orientations.", + "A MATLAB script was written to convert these files into orientation maps in the .ang format. This is the standard format used in TSL OIM Analysis by EDAX.", + "The microstructures simulated using DREAM.3D are spatially uniform, so using multiple sections is statistically equivalent to sampling larger sections, if the spacing between the sections is chosen to be larger than the maximum grain size to avoid sampling artefacts." + ], + "used_images": false + }, + { + "id": "pqac-bf758edb", + "context": "The excerpt describes an EBSD post-processing workflow where EBSD data was imported into the MTEX MATLAB library for analysis of a two-phase austenitic stainless steel. However, the excerpt provides minimal detail about MTEX's specific capabilities for grain reconstruction, texture analysis, or ODF calculations. The text mentions dataset cropping and segmentation into individual grains but does not elaborate on the technical implementations or comparative capabilities of MTEX versus other post-processing tools. The excerpt does not contain information about DREAM.3D, KAM (Kernel Average Misorientation), or peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, or PyMicro. While the paper addresses grain mapping and multiphase microstructure characterization, the specific EBSD post-processing methodologies and comparative software analysis requested are not covered in the provided text.", + "question": "Extract evidence relevant to comparing EBSD post-processing tools: (1) MTEX grain reconstruction and texture/ODF capabilities; (2) DREAM.3D grain segmentation and 2D/3D EBSD processing; (3) use of KAM or grain boundaries in EBSD workflows; (4) peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, PyMicro. Include exact capabilities and limitations where stated.", + "text": { + "embedding": null, + "text": "", + "name": "ball2023registrationbetweendct pages 2-4", + "media": [], + "doc": { + "embedding": null, + "docname": "ball2023registrationbetweendct", + "dockey": "0a036b5088d09190", + "citation": "J. Ball, J. Oddershede, C. Davis, C. Slater, M. Said, H. Vashishtha, \u0160. Michalik, David M. Collins School of Metallurgy, Materials, U. Birmingham, United Kingdom., Diamond Light Source Ltd., Xnovo Technology ApS, Denmark, Wmg, and U. Warwick. Registration between dct and ebsd datasets for multiphase microstructures. Materials Characterization, 204:113228, Oct 2023. URL: https://doi.org/10.1016/j.matchar.2023.113228, doi:10.1016/j.matchar.2023.113228.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "ball2023registrationbetweendct", + "bibtex": "@article{ball2023registrationbetweendct,\n author = \"Ball, J. and Oddershede, J. and Davis, C. and Slater, C. and Said, M. and Vashishtha, H. and Michalik, \u0160. and of Metallurgy, David M. Collins School and Materials and Birmingham, U. and Kingdom., United and Ltd., Diamond Light Source and ApS, Xnovo Technology and Denmark and Wmg and Warwick, U.\",\n title = \"Registration between DCT and EBSD datasets for multiphase microstructures\",\n year = \"2023\",\n journal = \"Materials Characterization\",\n volume = \"204\",\n pages = \"113228\",\n month = \"Oct\",\n doi = \"10.1016/j.matchar.2023.113228\",\n url = \"https://doi.org/10.1016/j.matchar.2023.113228\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n", + "authors": [ + "J. Ball", + "J. Oddershede", + "C. Davis", + "C. Slater", + "M. Said", + "H. Vashishtha", + "\u0160. Michalik", + "David M. Collins School of Metallurgy", + "Materials", + "U. Birmingham", + "United Kingdom.", + "Diamond Light Source Ltd.", + "Xnovo Technology ApS", + "Denmark", + "Wmg", + "U. Warwick" + ], + "publication_date": "2023-10-01T00:00:00Z", + "year": 2023, + "volume": "204", + "issue": null, + "issn": "1044-5803", + "pages": "113228", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2023.113228", + "title": "Registration between DCT and EBSD datasets for multiphase microstructures", + "citation_count": 8, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2023.113228", + "doi_url": "https://doi.org/10.1016/j.matchar.2023.113228", + "doc_id": "0a036b5088d09190", + "file_location": null, + "license": null, + "pdf_url": "https://ars.els-cdn.com/content/image/1-s2.0-S1044580323005879-ga1_lrg.jpg", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Ball, J. Oddershede, C. Davis, C. Slater, M. Said, H. Vashishtha, \u0160. Michalik, David M. Collins School of Metallurgy, Materials, U. Birmingham, United Kingdom., Diamond Light Source Ltd., Xnovo Technology ApS, Denmark, Wmg, and U. Warwick. Registration between dct and ebsd datasets for multiphase microstructures. Materials Characterization, 204:113228, Oct 2023. URL: https://doi.org/10.1016/j.matchar.2023.113228, doi:10.1016/j.matchar.2023.113228. This article has 8 citations and is from a peer-reviewed journal." + }, + "chunk_id": "99ea9a55-eb2f-4b25-90ec-3e5cf74d8d51" + }, + "score": 2, + "snippets": [ + "The EBSD map dataset was imported into the MTEX MATLAB library [35]." + ], + "used_images": false + }, + { + "id": "pqac-e9c242cb", + "context": "This excerpt focuses on DREAM.3D integration into the BisQue cloud platform for 3D microstructure reconstruction and analysis. It describes DREAM.3D capabilities including parameterization of reconstruction pipelines, high-performance cluster computing integration, and algorithms for 2D and 3D volume element sampling from large datasets. The paper reports findings about grain segmentation misorientation tolerances for additively manufactured materials and demonstrates that traditional 2D metallographic sectioning yields grain size measurements roughly half the size of equivalent 3D measurements in heterogeneous AM microstructures. However, the excerpt does not contain information about MTEX, orix, DefDAP, kikuchipy, pyxem, PyMicro, KAM, or comparative analysis of other EBSD post-processing tools.", + "question": "Extract evidence relevant to comparing EBSD post-processing tools: (1) MTEX grain reconstruction and texture/ODF capabilities; (2) DREAM.3D grain segmentation and 2D/3D EBSD processing; (3) use of KAM or grain boundaries in EBSD workflows; (4) peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, PyMicro. Include exact capabilities and limitations where stated.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 13-14", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "e9956fea-3d8b-49cb-817d-b77128e73ed8" + }, + "score": 3, + "snippets": [ + "The 3D microstructure reconstruction and analysis software DREAM.3D was integrated as a module into the BisQue cloud-based platform", + "Algorithms for the 2D and 3D volume element sampling from large 3D datasets", + "new insights into the choice of grain segmentation misorientation tolerances for additively manufactured materials were discovered", + "Using algorithms for the virtual 2D and 3D volume element sampling from large 3D datasets, traditional 2D metallographic sectioning was found to yield grain size measurements that were roughly half the size of the equivalent 3D measurements in heterogeneously distributed microstructures such as those found in AM materials" + ], + "used_images": false + }, + { + "id": "pqac-14206fb2", + "context": "This paper describes a stereological method applied to 2D EBSD orientation maps for obtaining grain boundary plane distributions (GBPD), using DREAM.3D for synthetic microstructure generation and simulation. The work demonstrates that stereological analysis is the fastest method for obtaining GBPDs but has limitations for textured materials due to biased viewing perspectives. The authors note that obtaining grain boundary character from 2D EBSD maps would be preferable to expensive 3D techniques like synchrotron X-ray diffraction or serial sectioning. However, the excerpt does not contain information about MTEX, KAM analysis, or the peer-reviewed software references mentioned (orix, DefDAP, kikuchipy, pyxem, PyMicro).", + "question": "Extract evidence relevant to comparing EBSD post-processing tools: (1) MTEX grain reconstruction and texture/ODF capabilities; (2) DREAM.3D grain segmentation and 2D/3D EBSD processing; (3) use of KAM or grain boundaries in EBSD workflows; (4) peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, PyMicro. Include exact capabilities and limitations where stated.", + "text": { + "embedding": null, + "text": "", + "name": "folwarczny2026accurategrainboundary pages 1-3", + "media": [], + "doc": { + "embedding": null, + "docname": "folwarczny2026accurategrainboundary", + "dockey": "59d8aa55a8c3757a", + "citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "folwarczny2026accurategrainboundary", + "bibtex": "@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n", + "authors": [ + "Martin Folwarczny", + "Ao Li", + "Rushvi Shah", + "Aaron Chote", + "Alexandra C. Austin", + "Yimin Zhu", + "Gregory S. Rohrer", + "Michael A. Jackson", + "Souhardh Kotakadi", + "Katharina Marquardt" + ], + "publication_date": "2026-02-01T00:00:00Z", + "year": 2026, + "volume": "280", + "issue": null, + "issn": "0304-3991", + "pages": "114262", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "title": "Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2025.114262", + "doi_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "doc_id": "59d8aa55a8c3757a", + "file_location": null, + "license": null, + "pdf_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal." + }, + "chunk_id": "4287fc76-69ba-485a-b51c-80b963650c1d" + }, + "score": 3, + "snippets": [ + "Stereology, applied to 2D EBSD orientation maps, is currently the fastest method of obtaining GBPDs.", + "We combined microstructure simulation of a textured material using DREAM.3D [25,26] with an analysis of these microstructures as if they were an experimental sample mapped by EBSD.", + "Being able to obtain accurate grain boundary character measurements using two-dimensional EBSD orientation maps would be a great simplification compared to three-dimensional orientation map techniques, such as synchrotron X-ray diffraction [19], serial sectioning [20\u201322] or transmission electron microscopy [23,24].", + "stereological analysis works well for samples with no preferred crystal orientations, i.e., non-textured materials", + "This stereological analysis works well for samples with no preferred crystal orientations, i.e., non-textured materials. This is because the analysis relies on supplied observations of individual types of grain boundaries from random observation perspectives, as it is a statistical method." + ], + "used_images": false + }, + { + "id": "pqac-193181fa", + "context": "The excerpt discusses DREAM.3D's capabilities for 3D EBSD data reconstruction and analysis. DREAM.3D creates a unified data structure for 3D EBSD data treatment and aggregates community-developed tools and algorithms. It operates as a pipeline with typically over a dozen steps, each with specific input variables and outputs. The paper describes integration of DREAM.3D into the BisQue cloud platform with parallelization capabilities. Key results include insights into grain segmentation misorientation tolerance in 3D EBSD datasets from TriBeam tomography of additively manufactured materials, and comparisons of grain size measurements between 2D metallographic slices and 3D DREAM.3D measures. The excerpt notes that reconstruction parameters (slice alignment, feature segmentation, artifact removal) significantly affect quantitative results extracted from 3D volumes. However, the excerpt does not contain information about MTEX, KAM, grain boundaries, or the Python-based tools (orix, DefDAP, kikuchipy, pyxem, PyMicro).", + "question": "Extract evidence relevant to comparing EBSD post-processing tools: (1) MTEX grain reconstruction and texture/ODF capabilities; (2) DREAM.3D grain segmentation and 2D/3D EBSD processing; (3) use of KAM or grain boundaries in EBSD workflows; (4) peer-reviewed software references for orix, DefDAP, kikuchipy, pyxem, PyMicro. Include exact capabilities and limitations where stated.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "e3d46ca7-0da2-467e-addc-e121db6ce344" + }, + "score": 4, + "snippets": [ + "Significant progress on the reconstruction and analysis of electron backscatter diffraction (EBSD) data has been made by a software package DREAM.3D [8], which creates a unified data structure for 3D data treatment and aggregates tools and algorithms collectively developed by the community.", + "In DREAM.3D, this process takes the form of a pipeline, which typically consists of well over a dozen steps, and in which each step has its own set of input variables and outputs.", + "A microstructure analytics and 3D reconstruction software package, DREAM.3D, was integrated as a module into a cloud-based platform, BisQue.", + "Parallelization of DREAM.3D module executions and the ability to parameterize pipeline variables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets", + "a comparison in grain size measurements was made between standard 2D metallographic slices and 3D measures using BisQue's parallelized DREAM.3D module executions.", + "Reconstruction of 3D datasets is a complex process with many degrees of freedom. The algorithms employed, and tolerances selected by the user to align slices, segment features, and remove data artifacts (i.e., remove noise), ultimately determine the quantitative data extracted from the volume." + ], + "used_images": false + }, + { + "id": "pqac-00000004", + "context": "The excerpt describes Dream3D's application to 2D/3D EBSD reconstruction from the Small IN100 dataset (117 slices of EBSD data with 189 x 201 scan points). Dream3D employs automated pipelines for grain boundary reconstruction, including full reconstruction, morphological statistics, crystallographic statistics, mesh generation, and grain boundary character distribution (GBCD) analysis. The segmentation achieved high performance metrics (average IoU of 91.58% and Dice score of 95.6%). However, limitations exist: the algorithm relies on Inverse Pole Figure (IPF) color contrast and can misidentify neighboring grains with similar IPF colors as the same grain, causing missegmentation. The excerpt acknowledges this as an inherent uncertainty when using this pipeline for grain segmentation. No specific information about misorientation thresholds is provided. The text notes that EBSD remains necessary when precise grain orientation information is required, suggesting Dream3D is best suited for rapidly quantifying grain size distribution with approximately 10% error tolerance.", + "question": "What does the literature say specifically about DREAM.3D grain segmentation algorithms, misorientation thresholds, 2D/3D EBSD reconstruction, automation/pipelines, and maturity?", + "text": { + "embedding": null, + "text": "", + "name": "yi2025rapidgrainsegmentation pages 4-7", + "media": [], + "doc": { + "embedding": null, + "docname": "yi2025rapidgrainsegmentation", + "dockey": "1d5a7c5c3c22ef5f", + "citation": "Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "yi2025rapidgrainsegmentation", + "bibtex": "@article{yi2025rapidgrainsegmentation,\n author = \"Yi, Yu-Tsen and Seo, Junwon and Murphy, Kevin and Rollett, Anthony D.\",\n title = \"Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach\",\n year = \"2025\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jan\",\n doi = \"10.1007/s40192-024-00390-2\",\n url = \"https://doi.org/10.1007/s40192-024-00390-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Yu-Tsen Yi", + "Junwon Seo", + "Kevin Murphy", + "Anthony D. Rollett" + ], + "publication_date": "2025-01-20T00:00:00Z", + "year": 2025, + "volume": null, + "issue": null, + "issn": "2193-9764", + "pages": null, + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-024-00390-2", + "title": "Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach", + "citation_count": 6, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-024-00390-2", + "doi_url": "https://doi.org/10.1007/s40192-024-00390-2", + "doc_id": "1d5a7c5c3c22ef5f", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-024-00390-2.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal." + }, + "chunk_id": "addb0754-f8ff-4723-b497-20c6cafae6d2" + }, + "score": 6, + "snippets": [ + "Running these pipelines sequentially produced a denoised 3D meshed grain boundary map. This study used 2D images of the six face labels as input for the pipeline, while the node-type images served as the ground truth.", + "The validation results obtained from the EBSD images and Dream3D are shown in Fig. 4, demonstrating an average Intersection over Union (IoU) and Dice score of 0.9158 and 0.956, respectively.", + "In some cases, the IPF colors of neighboring grains may be very similar, despite them being distinct grains. This lack of contrast can lead the algorithm to incorrectly treat them as the same grain, resulting in missegmentation.", + "The raw .ang files from the Small IN100 dataset were imported into Dream3D and processed using the following pipelines: (08) Small IN100 Full Reconstruction, (01) Small IN100 Morphological Statistics, (05) Small IN100 Crystallographic Statistics, (01) Small IN100 Quick Mesh, Small IN100 Smooth Mesh, (03) Small IN100 Mesh, and (04) Small IN100 GBCD" + ], + "used_images": false + }, + { + "id": "pqac-00000002", + "context": "The excerpt discusses DREAM.3D's integration into the BisQue cloud platform for 3D microstructure reconstruction. Key findings include: (1) grain segmentation misorientation tolerance optimization specifically for additively manufactured materials yielded new insights; (2) algorithms for 2D and 3D volume element sampling revealed that traditional 2D metallographic sectioning produces grain size measurements approximately half those of equivalent 3D measurements in heterogeneous AM microstructures; (3) the platform enables pipeline parameterization and parallelized module execution for reconstruction optimization; (4) automation includes massively parallelized slicing/cropping pipelines to simulate 2D versus 3D statistical grain measurements. The work demonstrates DREAM.3D's practical application for addressing discrepancies between 2D and 3D grain analysis in additively manufactured materials.", + "question": "What does the literature say specifically about DREAM.3D grain segmentation algorithms, misorientation thresholds, 2D/3D EBSD reconstruction, automation/pipelines, and maturity?", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 13-14", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "e9956fea-3d8b-49cb-817d-b77128e73ed8" + }, + "score": 7, + "snippets": [ + "new insights into the choice of grain segmentation misorientation tolerances for additively manufactured materials were discovered", + "Using algorithms for the virtual 2D and 3D volume element sampling from large 3D datasets, traditional 2D metallographic sectioning was found to yield grain size measurements that were roughly half the size of the equivalent 3D measurements in heterogeneously distributed microstructures such as those found in AM materials", + "The DREAM.3D pipeline parameterization combined with parallelized module executions has allowed for the exploration and optimization of dataset reconstructions", + "Algorithms for the 2D and 3D volume element sampling from large 3D datasets", + "These measurements were performed with a massively parallelized array of slicing/cropping pipelines to simulate the 2D vs 3D statistical grain measurements" + ], + "used_images": false + }, + { + "id": "pqac-00000003", + "context": "The excerpt describes DREAM.3D's application for 3D EBSD reconstruction and grain segmentation in additively manufactured materials. Grains are segmented using misorientation tolerance thresholds, with specific implementation details provided for 304L stainless steel (misorientation tolerance with minimum grain size of 64 voxels) and another alloy (2\u00b0 misorientation tolerance with 125 voxels minimum). The tool supports 2D EBSD slice reconstruction into coherent 3D volumes. Key automation features include parameterizable pipeline execution allowing parallel processing across multiple compute nodes with variable parameter ranges rather than discrete values. This parallelization capability reduces setup and analysis time for parameter studies. The excerpt indicates DREAM.3D was integrated with BisQue cloud infrastructure, suggesting mature implementation for workflow automation and data provenance tracking, though the excerpt does not extensively discuss algorithm maturity or theoretical foundations.", + "question": "What does the literature say specifically about DREAM.3D grain segmentation algorithms, misorientation thresholds, 2D/3D EBSD reconstruction, automation/pipelines, and maturity?", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 2-5", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "d3d6793f-4252-4fed-975f-f1cfef20256f" + }, + "score": 7, + "snippets": [ + "Grains were segmented using a misorientation tolerance as described in \"Misorientation Tolerance Parameterization for Feature Identification\"", + "grains containing less than 64 voxels were considered too small to be resolved and were removed", + "A DREAM.3D module was incorporated into BisQue to allow for the processing and treatment of 3D datasets", + "Anytime a processing pipeline is generated in DREAM.3D through BisQue, parameter input variables can be represented as a range of values (with sub-increments)", + "many independent instantiations of DREAM.3D can be run at once across multiple nodes on a compute cluster, dramatically decreasing setup and analysis time for detailed parameter studies", + "Individual slices were reconstructed into a coherent volume using DREAM.3D in BisQue" + ], + "used_images": false + }, + { + "id": "pqac-00000001", + "context": "The excerpt discusses DREAM.3D as a mature software package for 3D EBSD data reconstruction and microstructure analysis that aggregates community-developed algorithms. It operates as a multi-step pipeline with numerous input variables and outputs. Traditionally, DREAM.3D runs serially on workstations with predefined parameters. The authors integrated DREAM.3D into a cloud-based platform (BisQue) to enable parallelization and parameterization of pipeline variables across value ranges. This parallelization specifically provided insights about grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets of additively manufactured materials. The manual, serial approach previously limited exploration of the full parameter space due to long computation times scaling with dataset size and complexity, introducing bias in reconstruction processes.", + "question": "What does the literature say specifically about DREAM.3D grain segmentation algorithms, misorientation thresholds, 2D/3D EBSD reconstruction, automation/pipelines, and maturity?", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "e3d46ca7-0da2-467e-addc-e121db6ce344" + }, + "score": 8, + "snippets": [ + "Significant progress on the reconstruction and analysis of electron backscatter diffraction (EBSD) data has been made by a software package DREAM.3D [8], which creates a unified data structure for 3D data treatment and aggregates tools and algorithms collectively developed by the community.", + "In DREAM.3D, this process takes the form of a pipeline, which typically consists of well over a dozen steps, and in which each step has its own set of input variables and outputs.", + "DREAM.3D typically runs in a serial process on a workstation, with predefined input parameters.", + "Parallelization of DREAM.3D module executions and the ability to parameterize pipeline variables over a range of values have led to insights about the grain segmentation misorientation tolerance in TriBeam-collected 3D EBSD datasets", + "Due to the long computation times required to process 3D data, which often scales proportionately as dataset size and complexity increases, it is impractical to explore the full parameter space by manually running reconstructions." + ], + "used_images": false + }, + { + "id": "pqac-00000005", + "context": "The excerpt mentions DREAM.3D software as a computational tool that enables significant data evaluation and five-parameter grain boundary analysis, though it does not elaborate on specific segmentation algorithms. Regarding misorientation thresholds, the paper notes that typical values are 5\u00b0 in EDAX OIM software, with other choices possible. The work demonstrates critical differences between 2D and 3D EBSD segmentation: identical misorientation thresholds produce different results in 2D versus 3D cases because nonconvex grains may appear as separate grains in 2D slices but belong to single grains in 3D reconstruction. The paper emphasizes that proper segmentation attention is necessary to avoid misleading quantitative microstructure results. However, the excerpt provides limited information about automation/pipelines or the overall maturity level of these technologies.", + "question": "What does the literature say specifically about DREAM.3D grain segmentation algorithms, misorientation thresholds, 2D/3D EBSD reconstruction, automation/pipelines, and maturity?", + "text": { + "embedding": null, + "text": "", + "name": "stanek2020comparisonofsegmentation pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "stanek2020comparisonofsegmentation", + "dockey": "c163ded6c3506dd4", + "citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stanek2020comparisonofsegmentation", + "bibtex": "@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n", + "authors": [ + "J. Stanek", + "J. Kope\u010dek", + "P. Kr\u00e1l", + "I. Karafi\u00e1tov\u00e1", + "F. Seitl", + "V. Bene\u0161" + ], + "publication_date": "2020-01-01T00:00:00Z", + "year": 2020, + "volume": "58", + "issue": "05", + "issn": "0023-432X", + "pages": "301-319", + "journal": "Metallic Materials", + "publisher": "Central Library of the Slovak Academy of Sciences", + "url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=f5jvks8o3ufok0s0abmso3h0r0", + "title": "Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.4149/km_2020_5_301", + "doi_url": "https://doi.org/10.4149/km_2020_5_301", + "doc_id": "c163ded6c3506dd4", + "file_location": null, + "license": null, + "pdf_url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=buc29d44pbljhmfhjpava1l817", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "crossref", + "semantic_scholar", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations." + }, + "chunk_id": "a9d9080d-189f-4444-80b2-98aea1dd69c4" + }, + "score": 6, + "snippets": [ + "Computational tools such as DREAM 3D software enable significant data evaluation that is completed by a five-parameter analysis of grain boundaries.", + "The EBSD method considers a grain to be the set of all points (voxels or pixels, respectively) with a similar orientation, which are surrounded by a boundary of a given misorientation threshold. The typical threshold in the EDAX OIM software is 5\u00b0", + "The obtained segmentation into the grains depends on the chosen mis-orientation threshold, of course, but even if we use a fixed misorientation threshold, we may get different results of the segmentation in 2D and 3D", + "The differences between segmentation into grains in 2D and 3D are caused by the fact that in a 2D cutting plane, two sets of voxels may be surrounded by a boundary defined by a given threshold of misorientation so that the sets are detected as two grains, while in 3D case, all these points belong to one larger set" + ], + "used_images": false + }, + { + "id": "pqac-00000008", + "context": "# Summary\n\nThis excerpt discusses grain segmentation in EBSD measurements, comparing 2D and 3D approaches. The paper examines how grain segmentation depends on \"misorientation threshold\" definitions, noting that \"the typical threshold in the EDAX OIM software is 5\u00b0\" but other values are possible. The authors emphasize that \"even if we use a fixed misorientation threshold, we may get different results of the segmentation in 2D and 3D,\" creating challenges when comparing 2D and 3D EBSD results.\n\nThe excerpt mentions \"DREAM 3D software\" as a \"computational tool\" enabling \"significant data evaluation\" completed by \"five-parameter analysis of grain boundaries.\" However, the excerpt provides limited detail on specific grain reconstruction algorithms, texture analysis capabilities (ODF, pole figures), or misorientation-derived metrics like KAM and GND density. It focuses primarily on segmentation methodology rather than comprehensively comparing open-source Python packages for post-processing.", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "text": { + "embedding": null, + "text": "", + "name": "stanek2020comparisonofsegmentation pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "stanek2020comparisonofsegmentation", + "dockey": "c163ded6c3506dd4", + "citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stanek2020comparisonofsegmentation", + "bibtex": "@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n", + "authors": [ + "J. Stanek", + "J. Kope\u010dek", + "P. Kr\u00e1l", + "I. Karafi\u00e1tov\u00e1", + "F. Seitl", + "V. Bene\u0161" + ], + "publication_date": "2020-01-01T00:00:00Z", + "year": 2020, + "volume": "58", + "issue": "05", + "issn": "0023-432X", + "pages": "301-319", + "journal": "Metallic Materials", + "publisher": "Central Library of the Slovak Academy of Sciences", + "url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=f5jvks8o3ufok0s0abmso3h0r0", + "title": "Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.4149/km_2020_5_301", + "doi_url": "https://doi.org/10.4149/km_2020_5_301", + "doc_id": "c163ded6c3506dd4", + "file_location": null, + "license": null, + "pdf_url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=buc29d44pbljhmfhjpava1l817", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "crossref", + "semantic_scholar", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations." + }, + "chunk_id": "a9d9080d-189f-4444-80b2-98aea1dd69c4" + }, + "score": -1 + }, + { + "id": "pqac-00000009", + "context": "# Summary\n\nThis excerpt describes an automated MTEX-based procedure for reconstructing deformation twin hierarchies from EBSD datasets, particularly for heavily twinned microstructures (twin fractions >0.5). The paper demonstrates algorithms built on \"MTEX and MATLAB graph toolboxes\" that address challenges including missing twin relationships, grouping deformed grain fragments, identifying parent fragments, and handling multiple twin relationships.\n\nKey capabilities highlighted include: automated twin family tree reconstruction using a \"minimum spanning tree,\" identification of twin generations (demonstrated for third-generation twinning in \u03b1-Ti), and extraction of twin statistics at larger deformation levels. The work emphasizes MTEX's role in texture analysis and microstructure characterization but does not discuss Python-native alternatives or comparative analysis with other software packages for grain reconstruction or texture analysis.", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "text": { + "embedding": null, + "text": "", + "name": "savage2021anautomatedprocedure pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "savage2021anautomatedprocedure", + "dockey": "22c76dc107283440", + "citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "savage2021anautomatedprocedure", + "bibtex": "@article{savage2021anautomatedprocedure,\n author = \"Savage, Daniel J. and McCabe, Rodney J. and Knezevic, Marko\",\n title = \"An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\",\n volume = \"171\",\n issn = \"1044-5803\",\n url = \"https://doi.org/10.1016/j.matchar.2020.110808\",\n doi = \"10.1016/j.matchar.2020.110808\",\n journal = \"Materials Characterization\",\n publisher = \"Elsevier BV\",\n year = \"2021\",\n month = \"Jan\",\n pages = \"110808\"\n}\n", + "authors": [ + "Daniel J. Savage", + "Rodney J. McCabe", + "Marko Knezevic" + ], + "publication_date": "2021-01-01T00:00:00Z", + "year": 2021, + "volume": "171", + "issue": null, + "issn": "1044-5803", + "pages": "110808", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2020.110808", + "title": "An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures", + "citation_count": 20, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2020.110808", + "doi_url": "https://doi.org/10.1016/j.matchar.2020.110808", + "doc_id": "22c76dc107283440", + "file_location": null, + "license": null, + "pdf_url": "http://manuscript.elsevier.com/S1044580320322798/pdf/S1044580320322798.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal." + }, + "chunk_id": "2ffacaa6-d199-4e5b-b7cf-5b5bf864ee14" + }, + "score": -1 + }, + { + "id": "pqac-00000010", + "context": "# Summary\n\nThis excerpt describes DREAM.3D integration within BisQue, a cloud-based analysis platform for 3D EBSD datasets from additively manufactured materials. Key features include:\n\n**DREAM.3D Capabilities:**\n- Incorporated into BisQue for \"processing and treatment of 3D datasets\"\n- Enables parallelized processing across parameter spaces on compute clusters\n- Supports grain segmentation using \"misorientation tolerance\" thresholds\n- Reconstructs 3D volumes from serial-sectioned EBSD slices\n\n**Case Study (LENS 304L Stainless Steel):**\n- 200 GB dataset: 80 slices at 3 \u03bcm in-plane resolution, volume 816 \u00d7 834 \u00d7 240 \u03bcm\n- Confidence index (CI) \u2265 0.1 threshold (95% indexing accuracy)\n- Grain segmentation with minimum voxel requirements (64-125 voxels)\n- Equivalent grain diameter resolution of 14.9 \u03bcm\n\n**Infrastructure Advantage:**\n- Web-based visualization eliminates need for ParaView or HDFView, reducing hardware requirements for large datasets (tens of gigabytes)\n- Automatic data provenance tracking via unique URLs", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 2-5", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "d3d6793f-4252-4fed-975f-f1cfef20256f" + }, + "score": -1 + }, + { + "id": "pqac-00000007", + "context": "# Summary\n\nThis excerpt discusses grain segmentation methodology in EBSD data processing using the \"Segment Features (Misorientation)\" filter in DREAM.3D software. Key findings include:\n\nThe study compares 2D versus 3D grain segmentation approaches using simulated Laguerre tessellation data with \"1560 grains.\" Two material scenarios were tested: case (A) with low-angle grain boundaries (LAGB) and case (B) with high-angle grain boundaries (HAGB).\n\nCritical observation: \"the segmentations are entirely different\" when comparing 2D and 3D methods at misorientation thresholds of 2.5\u00b0 and 3\u00b0 in case (A), though \"for the threshold of 2\u00b0, we get nearly the same segmentation in both cases.\" The difference depends on specimen thickness\u20143D datasets with only \"a few slices\" show \"marginal\" differences from 2D segmentation.\n\nThe study varied misorientation thresholds (2\u00b0, 2.5\u00b0, 3\u00b0 for case A; 2\u00b0, 35\u00b0, 40\u00b0 for case B) and dataset thicknesses (10-100 slices) to systematically analyze segmentation behavior across dimensionalities.", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "text": { + "embedding": null, + "text": "", + "name": "stanek2020comparisonofsegmentation pages 2-7", + "media": [], + "doc": { + "embedding": null, + "docname": "stanek2020comparisonofsegmentation", + "dockey": "c163ded6c3506dd4", + "citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stanek2020comparisonofsegmentation", + "bibtex": "@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n", + "authors": [ + "J. Stanek", + "J. Kope\u010dek", + "P. Kr\u00e1l", + "I. Karafi\u00e1tov\u00e1", + "F. Seitl", + "V. Bene\u0161" + ], + "publication_date": "2020-01-01T00:00:00Z", + "year": 2020, + "volume": "58", + "issue": "05", + "issn": "0023-432X", + "pages": "301-319", + "journal": "Metallic Materials", + "publisher": "Central Library of the Slovak Academy of Sciences", + "url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=f5jvks8o3ufok0s0abmso3h0r0", + "title": "Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.4149/km_2020_5_301", + "doi_url": "https://doi.org/10.4149/km_2020_5_301", + "doc_id": "c163ded6c3506dd4", + "file_location": null, + "license": null, + "pdf_url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=buc29d44pbljhmfhjpava1l817", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "crossref", + "semantic_scholar", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations." + }, + "chunk_id": "01302fbe-a417-469d-bff7-fdede0f624e9" + }, + "score": -1 + }, + { + "id": "pqac-00000011", + "context": "# Summary\n\nThis excerpt discusses DREAM.3D, a \"microstructure analytics and 3D reconstruction software package\" that was integrated into BisQue, a cloud-based platform for materials science data analysis. The paper focuses on 3D EBSD reconstruction and grain segmentation from TriBeam-collected datasets of additively manufactured materials, exploring \"grain segmentation misorientation tolerance\" parameters. \n\nThe authors note that DREAM.3D \"typically runs in a serial process on a workstation, with predefined input parameters\" and describe limitations in current 3D microstructure analysis standardization. They highlight that DREAM.3D provides \"a unified data structure for 3D data treatment and aggregates tools and algorithms\" for EBSD data processing through pipeline-based workflows with \"well over a dozen steps.\"\n\nHowever, this excerpt does not directly address Python-native EBSD packages, their texture analysis capabilities (ODF, pole figures), misorientation metrics (KAM, GND), or comparisons with MTEX.", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "text": { + "embedding": null, + "text": "", + "name": "polonsky2019threedimensionalanalysisand pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + }, + "chunk_id": "e3d46ca7-0da2-467e-addc-e121db6ce344" + }, + "score": -1 + }, + { + "id": "pqac-00000006", + "context": "# Summary\n\nThis excerpt discusses a machine-learning approach called \"texture adaptive clustering and sampling\" (TACS) for generating representative Euler angle datasets from EBSD data. The study emphasizes that generating \"reduced-order, synthetic grain structure datasets that accurately represent the measured grain structure\" is important for crystal plasticity simulations while managing computational costs.\n\nKey capabilities mentioned include:\n- Creating representative Euler angle datasets that \"accurately mimic the crystallographic texture\"\n- Analyzing pole figures (PF) to characterize texture\n- TACS achieved Kolmogorov-Smirnov test scores \"nearing 0.9,\" compared to \"below 0.3\" for conventional methods\n- The approach is validated across \"twenty-two datasets, varying lattice structures, and complex crystallographic textures\"\n\nThe excerpt notes that **Dream3D** is used for grain morphology modeling and that machine learning methods have \"already been utilized to solve complex problems that are associated with EBSD datasets.\" However, this excerpt does not provide direct comparisons with specific Python-native packages (kikuchipy, orix, DefDAP, pyxem) or discuss grain reconstruction/segmentation, KAM, or GND density calculations needed for your question.", + "question": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "text": { + "embedding": null, + "text": "", + "name": "wanni2024machinelearningenhanced pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "wanni2024machinelearningenhanced", + "dockey": "bf3e624ed86e8966", + "citation": "J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "wanni2024machinelearningenhanced", + "bibtex": "@article{wanni2024machinelearningenhanced,\n author = \"Wanni, J. and Bronkhorst, C. A. and Thoma, D. J.\",\n title = \"Machine learning enhanced analysis of EBSD data for texture representation\",\n year = \"2024\",\n journal = \"npj Computational Materials\",\n volume = \"10\",\n pages = \"1-11\",\n month = \"Jun\",\n doi = \"10.1038/s41524-024-01324-4\",\n url = \"https://doi.org/10.1038/s41524-024-01324-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n", + "authors": [ + "J. Wanni", + "C. A. Bronkhorst", + "D. J. Thoma" + ], + "publication_date": "2024-06-26T00:00:00Z", + "year": 2024, + "volume": "10", + "issue": "1", + "issn": "2057-3960", + "pages": "1-11", + "journal": "npj Computational Materials", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41524-024-01324-4", + "title": "Machine learning enhanced analysis of EBSD data for texture representation", + "citation_count": 14, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41524-024-01324-4", + "doi_url": "https://doi.org/10.1038/s41524-024-01324-4", + "doc_id": "bf3e624ed86e8966", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41524-024-01324-4.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4. This article has 14 citations and is from a peer-reviewed journal." + }, + "chunk_id": "b10f0f18-25a1-41d9-a47c-d2abb61d0e19" + }, + "score": -1 + }, + { + "id": "pqac-00000012", + "context": "This peer-reviewed paper (Materials Characterization, 2021) demonstrates MTEX's capability for advanced EBSD analysis, specifically automated reconstruction of deformation twin hierarchies in heavily twinned microstructures. The work addresses critical challenges in twin identification including missing twin relationships, grouping of deformed grain fragments, and classification of multi-family twin relationships. The authors developed algorithms combining MTEX with MATLAB graph toolboxes to significantly improve automation compared to existing methodologies. The procedure successfully identifies third-generation twinning in \u03b1-titanium and can extract twin statistics at larger deformation levels than previously possible. However, the excerpt does not provide information about MTEX's performance regarding ODF, pole figures, or IPF analysis, nor does it discuss cautions regarding misorientation-threshold grain segmentation methods.", + "question": "What peer-reviewed evidence establishes MTEX as a benchmark for EBSD grain detection/reconstruction and crystallographic texture analysis, including ODF, pole figures, IPF, and advanced twin reconstruction? What cautions apply to misorientation-threshold grain segmentation?", + "text": { + "embedding": null, + "text": "", + "name": "savage2021anautomatedprocedure pages 1-2", + "media": [], + "doc": { + "embedding": null, + "docname": "savage2021anautomatedprocedure", + "dockey": "22c76dc107283440", + "citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "savage2021anautomatedprocedure", + "bibtex": "@article{savage2021anautomatedprocedure,\n author = \"Savage, Daniel J. and McCabe, Rodney J. and Knezevic, Marko\",\n title = \"An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\",\n volume = \"171\",\n issn = \"1044-5803\",\n url = \"https://doi.org/10.1016/j.matchar.2020.110808\",\n doi = \"10.1016/j.matchar.2020.110808\",\n journal = \"Materials Characterization\",\n publisher = \"Elsevier BV\",\n year = \"2021\",\n month = \"Jan\",\n pages = \"110808\"\n}\n", + "authors": [ + "Daniel J. Savage", + "Rodney J. McCabe", + "Marko Knezevic" + ], + "publication_date": "2021-01-01T00:00:00Z", + "year": 2021, + "volume": "171", + "issue": null, + "issn": "1044-5803", + "pages": "110808", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2020.110808", + "title": "An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures", + "citation_count": 20, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2020.110808", + "doi_url": "https://doi.org/10.1016/j.matchar.2020.110808", + "doc_id": "22c76dc107283440", + "file_location": null, + "license": null, + "pdf_url": "http://manuscript.elsevier.com/S1044580320322798/pdf/S1044580320322798.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal." + }, + "chunk_id": "2ffacaa6-d199-4e5b-b7cf-5b5bf864ee14" + }, + "score": 6, + "snippets": [ + "An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures", + "a set of algorithms built on the MTEX and MATLAB graph toolboxes for automatic reconstruction of deformation twin hierarchies from Electron Backscatter Diffraction (EBSD) datasets", + "The procedure is demonstrated to significantly improve the level of automation in reconstructing twin hierarchies in heavily twinned microstructure compared to other methodologies in literature" + ], + "used_images": false + }, + { + "id": "pqac-00000013", + "context": "This peer-reviewed article from Materials Characterization demonstrates MTEX's application for grain reconstruction and texture analysis in heavily twinned \u03b1-Ti microstructures. The authors use MTEX's texture analysis capabilities to handle complex microstructures with large twin volume fractions, multiple twin variants, and higher-order twinning. The paper presents inverse pole figure (IPF) EBSD maps and discusses texture analysis with strong c-axis texture observations. Regarding misorientation-threshold grain segmentation cautions, the authors note that due to strong texture, neighboring grains often have low-angle boundaries, and twins commonly chain across grain boundaries with only small misorientation differences. They address this by biasing reconstruction towards over-segmentation rather than under-segmentation and using different misorientation tolerances (1-2\u00b0) depending on loading direction. The authors acknowledge that under-segmentation and low-angle boundary treatment presents challenges in grain reconstruction.", + "question": "What peer-reviewed evidence establishes MTEX as a benchmark for EBSD grain detection/reconstruction and crystallographic texture analysis, including ODF, pole figures, IPF, and advanced twin reconstruction? What cautions apply to misorientation-threshold grain segmentation?", + "text": { + "embedding": null, + "text": "", + "name": "savage2021anautomatedprocedure pages 3-4", + "media": [], + "doc": { + "embedding": null, + "docname": "savage2021anautomatedprocedure", + "dockey": "22c76dc107283440", + "citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "savage2021anautomatedprocedure", + "bibtex": "@article{savage2021anautomatedprocedure,\n author = \"Savage, Daniel J. and McCabe, Rodney J. and Knezevic, Marko\",\n title = \"An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\",\n volume = \"171\",\n issn = \"1044-5803\",\n url = \"https://doi.org/10.1016/j.matchar.2020.110808\",\n doi = \"10.1016/j.matchar.2020.110808\",\n journal = \"Materials Characterization\",\n publisher = \"Elsevier BV\",\n year = \"2021\",\n month = \"Jan\",\n pages = \"110808\"\n}\n", + "authors": [ + "Daniel J. Savage", + "Rodney J. McCabe", + "Marko Knezevic" + ], + "publication_date": "2021-01-01T00:00:00Z", + "year": 2021, + "volume": "171", + "issue": null, + "issn": "1044-5803", + "pages": "110808", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2020.110808", + "title": "An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures", + "citation_count": 20, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2020.110808", + "doi_url": "https://doi.org/10.1016/j.matchar.2020.110808", + "doc_id": "22c76dc107283440", + "file_location": null, + "license": null, + "pdf_url": "http://manuscript.elsevier.com/S1044580320322798/pdf/S1044580320322798.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal." + }, + "chunk_id": "0692343a-b0a7-4f03-ad0a-4354d80a6a33" + }, + "score": 6, + "snippets": [ + "We leverage the flexibility and comprehensive texture analysis capabilities of MTEX and the MATLAB graph toolbox to address these shortcomings.", + "Grain fragments were reconstructed using the MTEX algorithm described in [55] and utilized a flood fill approach during grain reconstruction", + "MTEX is a comprehensive, versatile texture toolbox allowing several definitions of twin misorientations.", + "grains were reconstructed with a \u03b8frag = 1\u00b0 misorientation tolerance for ND and a \u03b8frag = 2\u00b0 misorientation tolerance for RD", + "Adding a boundary due to under segmentation is not easily done, and biasing grain fragment reconstruction towards over segmentation is advantageous." + ], + "used_images": false + } + ], + "references": "1. (folwarczny2026accurategrainboundary pages 5-9): Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal.\n\n2. (savage2021anautomatedprocedure pages 1-2): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\n3. (savage2021anautomatedprocedure pages 3-4): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\n4. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n5. (polonsky2019threedimensionalanalysisand pages 13-14): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n6. (polonsky2019threedimensionalanalysisand pages 2-5): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n7. (yi2025rapidgrainsegmentation pages 4-7): Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal.\n\n8. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\n9. (wanni2024machinelearningenhanced pages 1-2): J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4. This article has 14 citations and is from a peer-reviewed journal.\n\n10. (stanek2020comparisonofsegmentation pages 2-7): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.", + "formatted_answer": "Question: What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.\n\n# Open-source Python options for EBSD post-processing\n\n## Executive conclusion\n\nThere is currently **no single Python-native package that reproduces the full integrated scope and maturity of MTEX**. The most credible MATLAB-free solution is a composed stack:\n\n1. **kikuchipy + PyEBSDIndex** for pattern processing and indexing;\n2. **DREAM3D-NX through `simplnx`** for reproducible grain segmentation, cleanup, grain statistics, and especially 3D-capable workflows;\n3. **orix** as the crystallographic orientation, symmetry, misorientation, IPF, and pole-figure layer;\n4. optionally **DefDAP** for convenient 2D grain/deformation-map analysis;\n5. a separately tested module for **KAM**, and especially for **GND-density inversion**.\n\nFor the proposed automated AlSi10Mg pipeline, DREAM3D-NX/`simplnx` is the strongest grain-characterization backend, while orix is the best Python-native crystallographic foundation. kikuchipy itself is not a downstream replacement for MTEX, and pyxem is primarily a diffraction-microscopy/4D-STEM package rather than a conventional SEM-EBSD grain-analysis suite.\n\n| Package / ecosystem | Python-native status | Grain reconstruction / segmentation | Texture ODF / PF / IPF | KAM / GND | Maturity / reference status | Recommended role |\n|---|---|---|---|---|---|---|\n| MTEX | No; MATLAB-based | **Strong** for symmetry-aware 2D grain detection/reconstruction; advanced specialized workflows documented | **Strong**; benchmark for ODF, pole figures, IPF, texture components (folwarczny2026accurategrainboundary pages 5-9, savage2021anautomatedprocedure pages 1-2, savage2021anautomatedprocedure pages 3-4) | **Strong** in the MTEX ecosystem; widely treated as benchmark | Very mature, heavily cited benchmark; peer-reviewed advanced EBSD workflows documented (savage2021anautomatedprocedure pages 1-2, savage2021anautomatedprocedure pages 3-4) | Functional benchmark and validation target, but not MATLAB-free |\n| kikuchipy | Yes | **Absent / not core** for downstream grain reconstruction; primarily EBSD pattern processing/indexing support | **Partial / custom** via interoperability, not a full texture-analysis suite | **Absent / not core** | Open-source Python EBSD toolbox with literature record, but not a complete post-processing replacement in the retrieved evidence | Upstream pattern handling, preprocessing, indexing, data I/O; pair with orix and/or DREAM3D-NX |\n| orix | Yes | **Partial / custom**; orientation objects and misorientation logic support custom grain segmentation, but not MTEX-level turnkey reconstruction | **Strong / near-benchmark for orientation visualization**; IPF/PF native, ODF possible but typically less turnkey than MTEX | **Partial / custom**; KAM-like metrics and curvature workflows need user implementation/validation | Peer-reviewed library record available; positioned as crystallographic orientation library rather than full EBSD suite | Core Python orientation engine for custom EBSD post-processing and visualization |\n| DefDAP | Yes | **Partial**; useful 2D deformation/EBSD map grain analysis, but narrower than MTEX | **Partial**; some crystallographic plotting/analysis, not a full MTEX texture environment | **Partial**; deformation-oriented metrics are a strength, GND generally custom/limited | Open-source Python project with software-release citation; maturity moderate, community smaller than MTEX/DREAM3D | Good lightweight Python choice for 2D grain/deformation analysis after indexing |\n| DREAM3D-NX / simplnx | Yes (Python API around DREAM3D-NX) | **Strong**; mature pipeline-based segmentation/reconstruction, especially for 2D/3D EBSD and batch automation (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5, yi2025rapidgrainsegmentation pages 4-7, stanek2020comparisonofsegmentation pages 1-2) | **Partial**; some crystallographic statistics, but not as texture-centric or interactive as MTEX (yi2025rapidgrainsegmentation pages 4-7) | **Partial / custom**; not typically the first choice for KAM/GND-centric analysis | Mature EBSD/microstructure pipeline ecosystem with peer-reviewed reconstruction studies (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5) | Best Python-accessible option for robust automated grain segmentation/statistics in a MATLAB-free pipeline |\n| pyxem | Yes | **Absent / not core for SEM-EBSD**; focused on diffraction microscopy in TEM/4D-STEM, not a direct MTEX replacement | **Partial** for orientation mapping/visualization in diffraction-microscopy contexts | **Absent / not core** | Mature open-source microscopy library, but its center of gravity is 4D-STEM/SPED rather than conventional SEM-EBSD post-processing | Use only if your workflow spans diffraction microscopy; not recommended as primary EBSD grain-texture backend |\n| PyMicro | Yes | **Partial / custom**; useful microstructure data structures and crystallography, but less turnkey for EBSD grain reconstruction | **Partial / custom** | **Partial / custom** | Useful research software ecosystem, but less established as a mainstream EBSD post-processing stack in the retrieved evidence | Supporting library for custom workflows, data models, simulation coupling |\n| OpenXY / research scripts / pyEBSD-style code | Yes, but heterogeneous | **Partial / custom** to **strong in isolated tasks**; highly variable and often method-specific | **Partial / custom** | **Partial / custom**, including possible GND research implementations | Lowest standardization; often lacks broad peer-reviewed software validation or stable APIs | Fill specific gaps such as KAM/GND or custom segmentation, but validate carefully against MTEX/DREAM3D outputs |\n| Practical no-MATLAB stack for AM AlSi10Mg | Yes | **Strong overall** if composed as: kikuchipy/PyEBSDIndex + DREAM3D-NX for segmentation + orix/DefDAP for orientation/statistics | **Partial-to-strong overall**; strongest with orix plus custom ODF tooling | **Partial / custom overall**; KAM feasible, GND should be a separately validated module | Most realistic MATLAB-free path based on current open-source ecosystem and evidence balance (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5) | Recommended integrated pipeline rather than searching for a single Python package that fully replaces MTEX |\n\n\n*Table: This table compares major open-source Python-native or Python-accessible EBSD post-processing options against MTEX for grain reconstruction, texture analysis, and misorientation-derived metrics. It is designed to help choose a MATLAB-free downstream pipeline for automated EBSD workflows.*\n\n## The MTEX benchmark\n\nMTEX combines, in one environment, symmetry-aware EBSD import and manipulation, grain reconstruction, grain and boundary objects, texture estimation and plotting, misorientation statistics, KAM/GROD-type measures, and curvature/dislocation-density workflows. Its texture functionality includes ODF estimation, pole and inverse-pole figures, texture components, fibres, and quantitative texture measures. Published applications use MTEX for IPF maps, ODF-derived pole figures, grain-size and grain-shape statistics, and grain-boundary analysis. (folwarczny2026accurategrainboundary pages 5-9)\n\nMTEX also supports sophisticated extensions beyond simple threshold segmentation. A peer-reviewed workflow built on MTEX and MATLAB graph tools reconstructed deformation-twin hierarchies, grouped deformed fragments into parent families, treated missing relationships, and identified third-generation twins in heavily twinned \u03b1-Ti. (savage2021anautomatedprocedure pages 1-2) This illustrates the principal gap in Python: individual Python libraries provide many underlying operations, but MTEX supplies a more coherent grain/boundary/texture abstraction and a larger collection of specialized reconstruction workflows.\n\nMTEX is not immune to methodological choices. Strong textures and low-angle boundaries can cause under-segmentation or chains of crystallographically similar regions; one MTEX study deliberately biased reconstruction toward over-segmentation and used 1\u20132\u00b0 tolerances for its particular material. (savage2021anautomatedprocedure pages 3-4) More generally, grain counts and geometry depend strongly on the selected threshold, connectivity, cleanup, and minimum-grain-size rules. A conventional 5\u00b0 threshold is common but is not universally valid. (stanek2020comparisonofsegmentation pages 1-2)\n\n## Package assessment\n\n### 1. kikuchipy\n\n**Role.** kikuchipy is the natural upstream package in the proposed workflow. It is designed around EBSD patterns: vendor-file access, lazy handling of large pattern arrays, detector and projection geometry, background correction, pattern processing, simulation-related operations, and indexing interoperability. Its data model uses the HyperSpy ecosystem and orix crystallographic objects.\n\n**Grains and texture.** It should not be selected as the principal grain-reconstruction engine. Once indexing has produced orientation, phase, quality, and coordinate arrays, downstream segmentation and grain statistics generally require another package or custom code. Likewise, kikuchipy is not a comprehensive ODF/texture-analysis system and does not offer an MTEX-equivalent grain/boundary object model.\n\n**KAM/GND.** Neither is a core turnkey kikuchipy capability. KAM can be calculated from its output arrays with orix/NumPy-style operations, but masking, symmetry reduction, neighborhood definition, and exclusion of high-angle boundaries remain the pipeline developer\u2019s responsibility. GND inversion is substantially further outside its main scope.\n\n**Maturity.** kikuchipy is a well-developed, domain-specific open-source project with a software paper titled *kikuchipy: an open-source toolbox for analysis of electron backscatter diffraction patterns*. Its maturity is strongest for pattern-space operations, not for post-indexing microstructure analysis.\n\n### 2. orix\n\n**Role.** orix is the most important Python-native crystallographic foundation. It supplies rotations and orientations, crystal and specimen symmetry, fundamental-zone reduction, symmetry-aware misorientation, orientation colours, inverse-pole-figure colour keys, pole-figure-style projections, and orientation clustering. Its peer-reviewed methodological reference is Johnstone et al., *Density-based clustering of crystal (mis)orientations and the orix Python library*, *Journal of Applied Crystallography* 53 (2020), 1293\u20131298, DOI 10.1107/S1600576720011103.\n\n**Grain reconstruction.** orix contains the primitives needed to compare neighboring orientations and cluster orientation data, but it is not yet an MTEX-like turnkey EBSD topology package. A production grain segmenter still needs scan-grid connectivity, phase masks, confidence masks, flood-fill or graph labeling, small-region handling, and grain-property aggregation. Those steps can be implemented with NumPy, SciPy, scikit-image, xarray, and networkx, but then the laboratory owns validation and maintenance.\n\n**Texture.** orix is strong for IPF colouring, stereographic projections, symmetry handling, and discrete orientation distributions. It is suitable for PF/IPF visualization and custom orientation-density estimation. Nevertheless, MTEX remains substantially richer and more established for quantitative ODF estimation, kernel-selection conventions, harmonic representations, fibres/components, uncertainty-aware texture workflows, and publication-ready texture analysis. An ODF reconstructed from discrete EBSD orientations is also parameter-dependent; published work notes sensitivity to ODF computation settings in MTEX itself, so Python/MTEX agreement requires matching kernels, bandwidths, symmetry, weights, and normalization. A modern EBSD texture study used pole figures to characterize weak texture and illustrates the need to retain a statistically representative orientation population. (wanni2024machinelearningenhanced pages 1-2)\n\n**KAM/GND.** orix is a good basis for KAM because it provides symmetry-aware misorientation operations. It does not, by itself, turn KAM into a one-call validated EBSD workflow. GND requires spatial orientation gradients, reference-frame consistency, Burgers vectors/slip-system assumptions, Nye-tensor construction, and an underdetermined inversion; this remains custom research code rather than an MTEX-equivalent orix feature.\n\n### 3. DefDAP\n\n**Role.** DefDAP\u2014Deformation Data Analysis in Python\u2014is a Python package aimed at combining EBSD with deformation measurements such as high-resolution digital image correlation. Its citable software release is Atkinson et al., *DefDAP: Deformation Data Analysis in Python* (2020), DOI 10.5281/zenodo.3688097.\n\n**Grains.** DefDAP is useful for 2D EBSD maps, grain finding and grain-level analysis, crystallographic slip/twinning interpretation, boundary overlays, and linking deformation fields to grains. For a conventional single-phase 2D Al EBSD map it may be the quickest lightweight route from an orientation map to grains and grain-wise quantities.\n\n**Texture and misorientation metrics.** Its emphasis is local deformation and grain-scale mechanics, not comprehensive quantitative texture. It is less complete than MTEX for ODFs, texture components, and generalized boundary/reconstruction operations. KAM-like local misorientation calculations can be added or may be available through map-level operations depending on release, but GND-density inversion should not be assumed to be a complete, standardized capability without checking the exact version and validating the implementation.\n\n**Maturity.** DefDAP has been used in peer-reviewed deformation studies, but its community, documentation breadth, and EBSD feature coverage are smaller than MTEX or DREAM.3D. It is best regarded as a focused 2D deformation-analysis package rather than a universal EBSD backend.\n\n### 4. DREAM.3D and DREAM3D-NX/`simplnx`\n\n**Role and architecture.** Legacy DREAM.3D is an established C++/Qt microstructure-analysis platform. DREAM3D-NX is its current-generation successor; `simplnx` provides Python bindings and pipeline construction. Thus, it is **Python-accessible and scriptable**, although not pure Python internally. This distinction is normally beneficial for large EBSD maps because segmentation and feature-statistics filters execute in compiled code.\n\n**Grain reconstruction.** This is the strongest open-source alternative for automated grain segmentation. DREAM.3D uses pipeline filters for data import, masking, cleanup, symmetry-aware misorientation segmentation, feature identification, neighbor lists, grain morphology, crystallographic statistics, boundary meshes, and, where applicable, 3D reconstruction. Published work describes DREAM.3D as a unified data structure aggregating community-developed algorithms, with reconstruction pipelines commonly containing more than a dozen parameterized steps. (polonsky2019threedimensionalanalysisand pages 1-2)\n\nIts `Segment Features (Misorientation)` workflow has been used to compare 2D and 3D segmentation across several misorientation thresholds. (stanek2020comparisonofsegmentation pages 2-7) Published AM/TriBeam workflows reconstructed serial EBSD slices, segmented grains using explicit misorientation tolerances, removed features below specified voxel counts, filled gaps, and automated parameter sweeps. (polonsky2019threedimensionalanalysisand pages 2-5) Cloud integration further demonstrated parallel parameter studies, provenance tracking, and grain-size comparisons between 2D and 3D AM datasets. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14)\n\n**Texture.** DREAM.3D calculates crystallographic feature statistics and supports orientation-distribution data used in synthetic microstructure generation, but it is not as strong as MTEX for interactive quantitative texture analysis. PF/IPF maps and basic orientation statistics are feasible; sophisticated ODF reconstruction and texture-component analysis are better handled in orix plus custom numerical code\u2014or validated against MTEX. Published workflows frequently use DREAM.3D for microstructure generation/reconstruction and MTEX for detailed ODF/PF analysis, reflecting this division of strengths. (folwarczny2026accurategrainboundary pages 5-9)\n\n**KAM/GND.** DREAM3D-NX should not be selected primarily for KAM or GND. Depending on installed plugins and version, local orientation filters may cover some neighborhood measures, but these capabilities are less central and less transparent than its feature segmentation/statistics pipeline. A NumPy/orix KAM stage is easier to audit. GND remains a specialist calculation.\n\n**Cautions.** Segmentation is not parameter-free. The same threshold can produce different 2D and 3D connectivity, especially for nonconvex grains or paths connected outside the selected plane. (stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 2-7) Minimum feature size and cleanup can also shift the measured grain-size distribution. Consequently, pipeline JSON, threshold, neighborhood convention, phase mask, and deleted/fill operations should be retained as provenance.\n\n### 5. pyxem\n\npyxem is an open-source Python ecosystem for multidimensional diffraction microscopy, built around HyperSpy and commonly used for scanning/precession electron diffraction and 4D-STEM. It is valuable for virtual diffraction imaging, diffraction-vector analysis, orientation mapping, and related TEM workflows.\n\nIt is **not a direct downstream package for conventional SEM-EBSD orientation maps**. Grain segmentation from vendor-style EBSD grids, quantitative EBSD texture analysis, KAM, and GND are not its central abstractions. Its shared ecosystem with kikuchipy/orix can be useful if the project later combines EBSD with scanning electron diffraction, but pyxem should not be the primary AlSi10Mg grain-analysis backend.\n\n### 6. PyMicro\n\nPyMicro provides Python classes and algorithms for crystallography, microstructure representation, grains, orientations, meshes, and links to simulation and tomography. It can import or represent EBSD-like microstructures and is useful for custom data integration, synthetic microstructures, and computational-mechanics coupling.\n\nFor this task, its limitations are that conventional EBSD segmentation, comprehensive ODF analysis, KAM, and GND are not offered with the same turnkey breadth or community validation as MTEX. It is better viewed as an extensible microstructure framework than as a drop-in EBSD analysis application.\n\n### 7. OpenXY and other research code\n\nOpenXY is associated mainly with high-angular-resolution EBSD cross-correlation and strain/rotation analysis. It addresses a different problem from ordinary Hough-indexed orientation-map characterization and is not a complete grain/texture package. Some Python ports and research repositories calculate lattice curvature or GND density, but their conventions, slip-system bases, regularization, and maintenance vary.\n\nOther small packages carrying names such as `pyebsd`, custom scikit-image segmenters, or laboratory scripts can calculate IPF colours, KAM, or threshold-connected grains. They can be useful building blocks, but package names are sometimes ambiguous, releases may be unmaintained, and peer-reviewed software references are often absent. They should not be treated as equivalent to MTEX merely because they produce an IPF or KAM image.\n\n## Recommended pipeline for AM AlSi10Mg\n\n### Data contract after indexing\n\nExport or retain the following arrays and metadata:\n\n- quaternion or rotation matrix per scan point\u2014preferably not only Euler angles;\n- phase ID, indexed/unindexed mask, confidence or match score, and pattern quality;\n- physical `x`, `y` coordinates and step size;\n- scan shape and rectangular/hexagonal topology;\n- crystal symmetry and lattice parameters for fcc Al;\n- detector/sample/reference-frame conventions;\n- indexing dictionary or phase-library version.\n\nAlSi10Mg is commonly dominated in EBSD by fcc \u03b1-Al, while the fine Si network may be incompletely indexed at ordinary step sizes. Do not let unindexed Si-rich pixels automatically split Al grains: establish an explicit phase/mask and cleanup policy and report it.\n\n### Grain stage\n\nUse **DREAM3D-NX/`simplnx`** for the production pipeline:\n\n1. construct the image geometry and load quaternion, phase, mask, and quality arrays;\n2. remove only demonstrably unreliable points;\n3. segment same-phase connected pixels with a documented disorientation threshold;\n4. calculate feature centroids, equivalent diameters, areas, aspect ratios, mean orientations, neighbors, and boundary misorientations;\n5. retain both the raw feature labels and any post-cleanup labels;\n6. sweep plausible thresholds rather than fixing one unexamined value.\n\nFor AM AlSi10Mg, evaluate at least a low-angle-sensitive threshold and a conventional grain threshold\u2014for example, a controlled sweep around 2\u20135\u00b0\u2014because cellular/subgrain orientation gradients may be physically meaningful. This is not a recommendation to choose a universal value; published work demonstrates that changing tolerance can radically alter segmentation and that 2D and 3D results are not automatically equivalent. (polonsky2019threedimensionalanalysisand pages 2-5, stanek2020comparisonofsegmentation pages 1-2, stanek2020comparisonofsegmentation pages 2-7)\n\nIf only modest 2D maps are involved and direct Python object manipulation is more important than DREAM3D-NX pipelines, **DefDAP** is a reasonable alternative. A fully custom orix/scikit-image segmenter is defensible only if regression-tested against DREAM3D-NX or MTEX.\n\n### Texture stage\n\nUse **orix** for orientation conversion, symmetry reduction, IPF maps, pole figures, and discrete orientation statistics. For ODF estimation:\n\n- weight points or grains deliberately; pixel-weighted and one-orientation-per-grain ODFs answer different questions;\n- state the kernel and half-width/bandwidth;\n- enforce cubic crystal symmetry and the intended specimen symmetry;\n- normalize PF/ODF intensities consistently, usually in multiples of random distribution;\n- validate against a known texture or a small MTEX reference dataset.\n\nThis distinction matters in AM material: pixel weighting emphasizes large columnar grains, whereas grain weighting describes the population of grains. A recent study highlights the importance of preserving the full orientation distribution when reducing EBSD orientation datasets and uses PF maps to assess texture strength. (wanni2024machinelearningenhanced pages 1-2)\n\n### KAM\n\nKAM is straightforward to implement but easy to define inconsistently. For each indexed point, compute the mean symmetry-reduced disorientation to valid neighbors, while specifying:\n\n- first-, second-, or higher-order neighborhood;\n- whether diagonals are included;\n- maximum included disorientation, often chosen to exclude grain boundaries;\n- treatment of phase boundaries, map edges, and unindexed pixels;\n- degrees versus radians;\n- whether the map has been denoised before calculation.\n\nUse orix for disorientation and NumPy/SciPy for neighborhood operations. Store the number of valid neighbors along with KAM; otherwise values near pores, Si particles, and map boundaries are misleading. KAM should not be interpreted directly as dislocation density.\n\n### GND density\n\nGND is the least turnkey part of a MATLAB-free pipeline. A defensible calculation needs orientation gradients in physical units, lattice curvature, an explicit Nye-tensor convention, Burgers-vector/slip-system definitions, and a solution to an underdetermined inversion\u2014commonly L1 or L2 minimization. Report whether the result is total scalar GND density, a lower bound, or slip-system-resolved density.\n\nFor conventional EBSD, noise and angular resolution strongly affect spatial derivatives. The result also scales with step size and preprocessing. Therefore:\n\n1. preserve orientations at full precision;\n2. estimate angular noise on an undeformed/reference region;\n3. document derivative stencil and smoothing;\n4. perform step-size or downsampling sensitivity tests;\n5. benchmark synthetic rotation-gradient fields with known curvature;\n6. compare a representative map against an established MTEX or HR-EBSD implementation.\n\nNo named Python package in this comparison offers a broadly validated, turnkey GND workflow matching MTEX\u2019s integration. Treat GND as an independently versioned scientific module, not as a plotting option attached to the grain segmenter.\n\n## Practical ranking\n\n- **Best production grain segmentation and statistics:** DREAM3D-NX/`simplnx`.\n- **Best Python crystallographic foundation and PF/IPF layer:** orix.\n- **Best upstream pattern/indexing integration:** kikuchipy + PyEBSDIndex.\n- **Best lightweight 2D grain/deformation package:** DefDAP.\n- **Best for TEM diffraction-microscopy integration, not SEM-EBSD post-processing:** pyxem.\n- **Best custom microstructure/simulation framework:** PyMicro.\n- **No fully satisfactory turnkey Python winner for GND:** implement and validate separately.\n\nAccordingly, the recommended deployable stack is **kikuchipy/PyEBSDIndex \u2192 DREAM3D-NX/`simplnx` \u2192 orix-based texture and KAM \u2192 validated custom GND module**. This removes MATLAB from routine production while retaining MTEX only as an optional one-time validation reference. DREAM.3D\u2019s documented use on complex additively manufactured 3D EBSD datasets, including automated parameter studies and explicit misorientation segmentation, makes this architecture substantially lower-risk than building the entire downstream stage directly from NumPy/orix primitives. (polonsky2019threedimensionalanalysisand pages 1-2, polonsky2019threedimensionalanalysisand pages 13-14, polonsky2019threedimensionalanalysisand pages 2-5)\n\nReferences\n\n1. (folwarczny2026accurategrainboundary pages 5-9): Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal.\n\n2. (savage2021anautomatedprocedure pages 1-2): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\n3. (savage2021anautomatedprocedure pages 3-4): Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal.\n\n4. (polonsky2019threedimensionalanalysisand pages 1-2): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n5. (polonsky2019threedimensionalanalysisand pages 13-14): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n6. (polonsky2019threedimensionalanalysisand pages 2-5): Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal.\n\n7. (yi2025rapidgrainsegmentation pages 4-7): Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal.\n\n8. (stanek2020comparisonofsegmentation pages 1-2): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.\n\n9. (wanni2024machinelearningenhanced pages 1-2): J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4. This article has 14 citations and is from a peer-reviewed journal.\n\n10. (stanek2020comparisonofsegmentation pages 2-7): J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations.", + "graded_answer": null, + "cost": 0.0, + "token_counts": {}, + "config_md5": "44fbe82f4172525021b58c01857cebcc", + "tool_history": [ + [ + "create_plan" + ], + [ + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search" + ], + [ + "gather_evidence" + ], + [ + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search", + "paper_search" + ], + [ + "gather_evidence" + ], + [ + "gather_evidence", + "read_text" + ], + [ + "update_plan" + ], + [ + "gather_evidence" + ], + [ + "update_plan" + ], + [ + "update_plan" + ], + [ + "update_plan" + ], + [ + "create_artifact" + ], + [ + "update_plan" + ], + [ + "update_plan" + ], + [ + "answer" + ] + ], + "citeable_context_ids": [ + "pqac-00000000", + "pqac-00000001", + "pqac-00000002", + "pqac-00000003", + "pqac-00000004", + "pqac-00000005", + "pqac-00000006", + "pqac-00000007", + "pqac-00000008", + "pqac-00000009", + "pqac-00000010", + "pqac-00000011", + "pqac-00000012", + "pqac-00000013" + ], + "artifacts": { + "artifact-00": "| Package / ecosystem | Python-native status | Grain reconstruction / segmentation | Texture ODF / PF / IPF | KAM / GND | Maturity / reference status | Recommended role |\n|---|---|---|---|---|---|---|\n| MTEX | No; MATLAB-based | **Strong** for symmetry-aware 2D grain detection/reconstruction; advanced specialized workflows documented | **Strong**; benchmark for ODF, pole figures, IPF, texture components (pqac-00000000, pqac-00000012, pqac-00000013) | **Strong** in the MTEX ecosystem; widely treated as benchmark | Very mature, heavily cited benchmark; peer-reviewed advanced EBSD workflows documented (pqac-00000012, pqac-00000013) | Functional benchmark and validation target, but not MATLAB-free |\n| kikuchipy | Yes | **Absent / not core** for downstream grain reconstruction; primarily EBSD pattern processing/indexing support | **Partial / custom** via interoperability, not a full texture-analysis suite | **Absent / not core** | Open-source Python EBSD toolbox with literature record, but not a complete post-processing replacement in the retrieved evidence | Upstream pattern handling, preprocessing, indexing, data I/O; pair with orix and/or DREAM3D-NX |\n| orix | Yes | **Partial / custom**; orientation objects and misorientation logic support custom grain segmentation, but not MTEX-level turnkey reconstruction | **Strong / near-benchmark for orientation visualization**; IPF/PF native, ODF possible but typically less turnkey than MTEX | **Partial / custom**; KAM-like metrics and curvature workflows need user implementation/validation | Peer-reviewed library record available; positioned as crystallographic orientation library rather than full EBSD suite | Core Python orientation engine for custom EBSD post-processing and visualization |\n| DefDAP | Yes | **Partial**; useful 2D deformation/EBSD map grain analysis, but narrower than MTEX | **Partial**; some crystallographic plotting/analysis, not a full MTEX texture environment | **Partial**; deformation-oriented metrics are a strength, GND generally custom/limited | Open-source Python project with software-release citation; maturity moderate, community smaller than MTEX/DREAM3D | Good lightweight Python choice for 2D grain/deformation analysis after indexing |\n| DREAM3D-NX / simplnx | Yes (Python API around DREAM3D-NX) | **Strong**; mature pipeline-based segmentation/reconstruction, especially for 2D/3D EBSD and batch automation (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000004, pqac-00000005, pqac-00000011) | **Partial**; some crystallographic statistics, but not as texture-centric or interactive as MTEX (pqac-00000004) | **Partial / custom**; not typically the first choice for KAM/GND-centric analysis | Mature EBSD/microstructure pipeline ecosystem with peer-reviewed reconstruction studies (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Best Python-accessible option for robust automated grain segmentation/statistics in a MATLAB-free pipeline |\n| pyxem | Yes | **Absent / not core for SEM-EBSD**; focused on diffraction microscopy in TEM/4D-STEM, not a direct MTEX replacement | **Partial** for orientation mapping/visualization in diffraction-microscopy contexts | **Absent / not core** | Mature open-source microscopy library, but its center of gravity is 4D-STEM/SPED rather than conventional SEM-EBSD post-processing | Use only if your workflow spans diffraction microscopy; not recommended as primary EBSD grain-texture backend |\n| PyMicro | Yes | **Partial / custom**; useful microstructure data structures and crystallography, but less turnkey for EBSD grain reconstruction | **Partial / custom** | **Partial / custom** | Useful research software ecosystem, but less established as a mainstream EBSD post-processing stack in the retrieved evidence | Supporting library for custom workflows, data models, simulation coupling |\n| OpenXY / research scripts / pyEBSD-style code | Yes, but heterogeneous | **Partial / custom** to **strong in isolated tasks**; highly variable and often method-specific | **Partial / custom** | **Partial / custom**, including possible GND research implementations | Lowest standardization; often lacks broad peer-reviewed software validation or stable APIs | Fill specific gaps such as KAM/GND or custom segmentation, but validate carefully against MTEX/DREAM3D outputs |\n| Practical no-MATLAB stack for AM AlSi10Mg | Yes | **Strong overall** if composed as: kikuchipy/PyEBSDIndex + DREAM3D-NX for segmentation + orix/DefDAP for orientation/statistics | **Partial-to-strong overall**; strongest with orix plus custom ODF tooling | **Partial / custom overall**; KAM feasible, GND should be a separately validated module | Most realistic MATLAB-free path based on current open-source ecosystem and evidence balance (pqac-00000001, pqac-00000002, pqac-00000003, pqac-00000011) | Recommended integrated pipeline rather than searching for a single Python package that fully replaces MTEX |\n\n\n*Table: This table compares major open-source Python-native or Python-accessible EBSD post-processing options against MTEX for grain reconstruction, texture analysis, and misorientation-derived metrics. It is designed to help choose a MATLAB-free downstream pipeline for automated EBSD workflows.*" + }, + "candidates": { + "call_kcmzfiVWrTdIWQyBi42VuIon": [], + "call_VIQzh60P2I1sB79k1x1yJiua": [ + { + "embedding": null, + "docname": "bachmann2022efficientreconstructionof", + "dockey": "b15a2bca44cba66a", + "citation": "Bj\u00f6rn-Ivo Bachmann, Martin M\u00fcller, Dominik Britz, Ali Riza Durmaz, Marc Ackermann, Oleg Shchyglo, Thorsten Staudt, and Frank M\u00fccklich. Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy. Frontiers in Materials, Oct 2022. URL: https://doi.org/10.3389/fmats.2022.1033505, doi:10.3389/fmats.2022.1033505.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "bachmann2022efficientreconstructionof", + "bibtex": "@article{bachmann2022efficientreconstructionof,\n author = \"Bachmann, Bj\u00f6rn-Ivo and M\u00fcller, Martin and Britz, Dominik and Durmaz, Ali Riza and Ackermann, Marc and Shchyglo, Oleg and Staudt, Thorsten and M\u00fccklich, Frank\",\n title = \"Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy\",\n year = \"2022\",\n journal = \"Frontiers in Materials\",\n volume = \"9\",\n month = \"Oct\",\n doi = \"10.3389/fmats.2022.1033505\",\n url = \"https://doi.org/10.3389/fmats.2022.1033505\",\n publisher = \"Frontiers Media SA\",\n issn = \"2296-8016\"\n}\n", + "authors": [ + "Bj\u00f6rn-Ivo Bachmann", + "Martin M\u00fcller", + "Dominik Britz", + "Ali Riza Durmaz", + "Marc Ackermann", + "Oleg Shchyglo", + "Thorsten Staudt", + "Frank M\u00fccklich" + ], + "publication_date": "2022-10-11T00:00:00Z", + "year": 2022, + "volume": "9", + "issue": null, + "issn": "2296-8016", + "pages": null, + "journal": "Frontiers in Materials", + "publisher": "Frontiers Media SA", + "url": "https://doi.org/10.3389/fmats.2022.1033505", + "title": "Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy", + "citation_count": 19, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.3389/fmats.2022.1033505", + "doi_url": "https://doi.org/10.3389/fmats.2022.1033505", + "doc_id": "b15a2bca44cba66a", + "file_location": null, + "license": null, + "pdf_url": "https://www.frontiersin.org/articles/10.3389/fmats.2022.1033505/pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Bj\u00f6rn-Ivo Bachmann, Martin M\u00fcller, Dominik Britz, Ali Riza Durmaz, Marc Ackermann, Oleg Shchyglo, Thorsten Staudt, and Frank M\u00fccklich. Efficient reconstruction of prior austenite grains in steel from etched light optical micrographs using deep learning and annotations from correlative microscopy. Frontiers in Materials, Oct 2022. URL: https://doi.org/10.3389/fmats.2022.1033505, doi:10.3389/fmats.2022.1033505. This article has 19 citations." + } + ], + "call_26RNCIR7p5ooimL1oHJKDxPe": [], + "call_TWgXFKMItXYFpWEbgg83FfU9": [ + { + "embedding": null, + "docname": "murgas2024modelingcomplexpolycrystalline", + "dockey": "c8c5ee9faa706563", + "citation": "Brayan Murgas, Joshua Stickel, Luke Brewer, and Somnath Ghosh. Modeling complex polycrystalline alloys using a generative adversarial network enabled computational platform. Nature Communications, Nov 2024. URL: https://doi.org/10.1038/s41467-024-53865-3, doi:10.1038/s41467-024-53865-3.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "murgas2024modelingcomplexpolycrystalline", + "bibtex": "@article{murgas2024modelingcomplexpolycrystalline,\n author = \"Murgas, Brayan and Stickel, Joshua and Brewer, Luke and Ghosh, Somnath\",\n title = \"Modeling complex polycrystalline alloys using a Generative Adversarial Network enabled computational platform\",\n year = \"2024\",\n journal = \"Nature Communications\",\n volume = \"15\",\n month = \"Nov\",\n doi = \"10.1038/s41467-024-53865-3\",\n url = \"https://doi.org/10.1038/s41467-024-53865-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2041-1723\"\n}\n", + "authors": [ + "Brayan Murgas", + "Joshua Stickel", + "Luke Brewer", + "Somnath Ghosh" + ], + "publication_date": "2024-11-01T00:00:00Z", + "year": 2024, + "volume": "15", + "issue": "1", + "issn": "2041-1723", + "pages": null, + "journal": "Nature Communications", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41467-024-53865-3", + "title": "Modeling complex polycrystalline alloys using a Generative Adversarial Network enabled computational platform", + "citation_count": 23, + "bibtex_type": "article", + "source_quality": 3, + "is_retracted": null, + "doi": "10.1038/s41467-024-53865-3", + "doi_url": "https://doi.org/10.1038/s41467-024-53865-3", + "doc_id": "c8c5ee9faa706563", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41467-024-53865-3.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Brayan Murgas, Joshua Stickel, Luke Brewer, and Somnath Ghosh. Modeling complex polycrystalline alloys using a generative adversarial network enabled computational platform. Nature Communications, Nov 2024. URL: https://doi.org/10.1038/s41467-024-53865-3, doi:10.1038/s41467-024-53865-3. This article has 23 citations and is from a highest quality peer-reviewed journal." + }, + { + "embedding": null, + "docname": "yi2025rapidgrainsegmentation", + "dockey": "1d5a7c5c3c22ef5f", + "citation": "Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "yi2025rapidgrainsegmentation", + "bibtex": "@article{yi2025rapidgrainsegmentation,\n author = \"Yi, Yu-Tsen and Seo, Junwon and Murphy, Kevin and Rollett, Anthony D.\",\n title = \"Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach\",\n year = \"2025\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n month = \"Jan\",\n doi = \"10.1007/s40192-024-00390-2\",\n url = \"https://doi.org/10.1007/s40192-024-00390-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Yu-Tsen Yi", + "Junwon Seo", + "Kevin Murphy", + "Anthony D. Rollett" + ], + "publication_date": "2025-01-20T00:00:00Z", + "year": 2025, + "volume": null, + "issue": null, + "issn": "2193-9764", + "pages": null, + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-024-00390-2", + "title": "Rapid Grain Segmentation of Heat-treated and Annealed LPBF Haynes 282 Using an Unsupervised Learning-Based Computer Vision Approach", + "citation_count": 6, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-024-00390-2", + "doi_url": "https://doi.org/10.1007/s40192-024-00390-2", + "doc_id": "1d5a7c5c3c22ef5f", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-024-00390-2.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Yu-Tsen Yi, Junwon Seo, Kevin Murphy, and Anthony D. Rollett. Rapid grain segmentation of heat-treated and annealed lpbf haynes 282 using an unsupervised learning-based computer vision approach. Integrating Materials and Manufacturing Innovation, Jan 2025. URL: https://doi.org/10.1007/s40192-024-00390-2, doi:10.1007/s40192-024-00390-2. This article has 6 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "stanek2020comparisonofsegmentation", + "dockey": "c163ded6c3506dd4", + "citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "stanek2020comparisonofsegmentation", + "bibtex": "@article{stanek2020comparisonofsegmentation,\n author = \"Stanek, J. and Kope\u010dek, J. and Kr\u00e1l, P. and Karafi\u00e1tov\u00e1, I. and Seitl, F. and Bene\u0161, V.\",\n title = \"Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials\",\n year = \"2020\",\n journal = \"Metallic Materials\",\n volume = \"58\",\n pages = \"301-319\",\n month = \"Jan\",\n doi = \"10.4149/km\\_2020\\_5\\_301\",\n url = \"https://doi.org/10.4149/km\\_2020\\_5\\_301\",\n publisher = \"Central Library of the Slovak Academy of Sciences\",\n issue = \"05\",\n issn = \"0023-432X\"\n}\n", + "authors": [ + "J. Stanek", + "J. Kope\u010dek", + "P. Kr\u00e1l", + "I. Karafi\u00e1tov\u00e1", + "F. Seitl", + "V. Bene\u0161" + ], + "publication_date": "2020-01-01T00:00:00Z", + "year": 2020, + "volume": "58", + "issue": "05", + "issn": "0023-432X", + "pages": "301-319", + "journal": "Metallic Materials", + "publisher": "Central Library of the Slovak Academy of Sciences", + "url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=f5jvks8o3ufok0s0abmso3h0r0", + "title": "Comparison of segmentation of 2D and 3D EBSD measurements in polycrystalline materials", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.4149/km_2020_5_301", + "doi_url": "https://doi.org/10.4149/km_2020_5_301", + "doc_id": "c163ded6c3506dd4", + "file_location": null, + "license": null, + "pdf_url": "http://www.elis.sk/download_file.php?product_id=6891&session_id=buc29d44pbljhmfhjpava1l817", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "crossref", + "semantic_scholar", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Stanek, J. Kope\u010dek, P. Kr\u00e1l, I. Karafi\u00e1tov\u00e1, F. Seitl, and V. Bene\u0161. Comparison of segmentation of 2d and 3d ebsd measurements in polycrystalline materials. Metallic Materials, 58:301-319, Jan 2020. URL: https://doi.org/10.4149/km\\_2020\\_5\\_301, doi:10.4149/km\\_2020\\_5\\_301. This article has 4 citations." + }, + { + "embedding": null, + "docname": "folwarczny2026accurategrainboundary", + "dockey": "59d8aa55a8c3757a", + "citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "folwarczny2026accurategrainboundary", + "bibtex": "@article{folwarczny2026accurategrainboundary,\n author = \"Folwarczny, Martin and Li, Ao and Shah, Rushvi and Chote, Aaron and Austin, Alexandra C. and Zhu, Yimin and Rohrer, Gregory S. and Jackson, Michael A. and Kotakadi, Souhardh and Marquardt, Katharina\",\n title = \"Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps\",\n volume = \"280\",\n issn = \"0304-3991\",\n url = \"https://doi.org/10.1016/j.ultramic.2025.114262\",\n doi = \"10.1016/j.ultramic.2025.114262\",\n journal = \"Ultramicroscopy\",\n publisher = \"Elsevier BV\",\n year = \"2026\",\n month = \"Feb\",\n pages = \"114262\"\n}\n", + "authors": [ + "Martin Folwarczny", + "Ao Li", + "Rushvi Shah", + "Aaron Chote", + "Alexandra C. Austin", + "Yimin Zhu", + "Gregory S. Rohrer", + "Michael A. Jackson", + "Souhardh Kotakadi", + "Katharina Marquardt" + ], + "publication_date": "2026-02-01T00:00:00Z", + "year": 2026, + "volume": "280", + "issue": null, + "issn": "0304-3991", + "pages": "114262", + "journal": "Ultramicroscopy", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "title": "Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.ultramic.2025.114262", + "doi_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "doc_id": "59d8aa55a8c3757a", + "file_location": null, + "license": null, + "pdf_url": "https://doi.org/10.1016/j.ultramic.2025.114262", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Martin Folwarczny, Ao Li, Rushvi Shah, Aaron Chote, Alexandra C. Austin, Yimin Zhu, Gregory S. Rohrer, Michael A. Jackson, Souhardh Kotakadi, and Katharina Marquardt. Accurate grain boundary plane distributions for textured microstructures from stereological analysis of orthogonal two-dimensional electron backscatter diffraction orientation maps. Ultramicroscopy, 280:114262, Feb 2026. URL: https://doi.org/10.1016/j.ultramic.2025.114262, doi:10.1016/j.ultramic.2025.114262. This article has 0 citations and is from a peer-reviewed journal." + } + ], + "call_oB8LA7Kpebnw5E9YNkkGkAve": [ + { + "embedding": null, + "docname": "heinig2022highresolutioncrystal", + "dockey": "2704548c22fe37a7", + "citation": "Mario F. Heinig, Dipanwita Chatterjee, Antonius T.J. van Helvoort, Jakob Birkedal Wagner, Shima Kadkhodazadeh, H\u00e5kon Wiik \u00c5nes, Frank Niessen, and Alice Bastos da Silva Fanta. High resolution crystal orientation mapping of ultrathin films in sem and tem. Jul 2022. URL: https://doi.org/10.1016/j.matchar.2022.111931, doi:10.1016/j.matchar.2022.111931.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "heinig2022highresolutioncrystal", + "bibtex": "@misc{heinig2022highresolutioncrystal,\n author = \"Heinig, Mario F. and Chatterjee, Dipanwita and van Helvoort, Antonius T.J. and Wagner, Jakob Birkedal and Kadkhodazadeh, Shima and \u00c5nes, H\u00e5kon Wiik and Niessen, Frank and da Silva Fanta, Alice Bastos\",\n title = \"High resolution crystal orientation mapping of ultrathin films in SEM and TEM\",\n year = \"2022\",\n journal = \"Materials Characterization\",\n volume = \"189\",\n pages = \"111931\",\n month = \"Jul\",\n doi = \"10.1016/j.matchar.2022.111931\",\n url = \"https://doi.org/10.1016/j.matchar.2022.111931\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n", + "authors": [ + "Mario F. Heinig", + "Dipanwita Chatterjee", + "Antonius T.J. van Helvoort", + "Jakob Birkedal Wagner", + "Shima Kadkhodazadeh", + "H\u00e5kon Wiik \u00c5nes", + "Frank Niessen", + "Alice Bastos da Silva Fanta" + ], + "publication_date": "2022-07-01T00:00:00Z", + "year": 2022, + "volume": "189", + "issue": null, + "issn": "1044-5803", + "pages": "111931", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2022.111931", + "title": "High resolution crystal orientation mapping of ultrathin films in SEM and TEM", + "citation_count": 6, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2022.111931", + "doi_url": "https://doi.org/10.1016/j.matchar.2022.111931", + "doc_id": "2704548c22fe37a7", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/pii/S1044580322002133", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Mario F. Heinig, Dipanwita Chatterjee, Antonius T.J. van Helvoort, Jakob Birkedal Wagner, Shima Kadkhodazadeh, H\u00e5kon Wiik \u00c5nes, Frank Niessen, and Alice Bastos da Silva Fanta. High resolution crystal orientation mapping of ultrathin films in sem and tem. Jul 2022. URL: https://doi.org/10.1016/j.matchar.2022.111931, doi:10.1016/j.matchar.2022.111931. This article has 6 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "ophus2022automatedcrystalorientation", + "dockey": "253bbe76737c510e", + "citation": "Colin Ophus, Steven E Zeltmann, Alexandra Bruefach, Alexander Rakowski, Benjamin H Savitzky, Andrew M Minor, and MC Scott. Automated crystal orientation mapping in py4dstem using sparse correlation matching. Microscopy and Microanalysis, 28:390-403, Oct 2022. URL: https://doi.org/10.48550/arxiv.2111.00171, doi:10.48550/arxiv.2111.00171.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "ophus2022automatedcrystalorientation", + "bibtex": "@article{ophus2022automatedcrystalorientation,\n author = \"Ophus, Colin and Zeltmann, Steven E and Bruefach, Alexandra and Rakowski, Alexander and Savitzky, Benjamin H and Minor, Andrew M and Scott, MC\",\n title = \"Automated Crystal Orientation Mapping in py4DSTEM using Sparse Correlation Matching\",\n year = \"2022\",\n journal = \"Microscopy and Microanalysis\",\n volume = \"28\",\n pages = \"390-403\",\n month = \"Oct\",\n doi = \"10.48550/arxiv.2111.00171\",\n url = \"https://doi.org/10.48550/arxiv.2111.00171\",\n publisher = \"arXiv\"\n}\n", + "authors": [ + "Colin Ophus", + "Steven E Zeltmann", + "Alexandra Bruefach", + "Alexander Rakowski", + "Benjamin H Savitzky", + "Andrew M Minor", + "MC Scott" + ], + "publication_date": "2021-10-30T00:00:00Z", + "year": 2022, + "volume": "28", + "issue": null, + "issn": null, + "pages": "390-403", + "journal": "Microscopy and Microanalysis", + "publisher": "arXiv", + "url": "https://arxiv.org/abs/2111.00171", + "title": "Automated Crystal Orientation Mapping in py4DSTEM using Sparse Correlation Matching", + "citation_count": 115, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.48550/arxiv.2111.00171", + "doi_url": "https://doi.org/10.48550/arxiv.2111.00171", + "doc_id": "253bbe76737c510e", + "file_location": null, + "license": null, + "pdf_url": "https://www.cambridge.org/core/services/aop-cambridge-core/content/view/AFE4FE01718EA756F4E548B223F4E420/S1431927622000101a.pdf/automated_crystal_orientation_mapping_in_py4dstem_using_sparse_correlation_matching.pdf", + "other": { + "client_source": [ + "paper_search", + "datacite", + "semantic_scholar", + "datacite", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Colin Ophus, Steven E Zeltmann, Alexandra Bruefach, Alexander Rakowski, Benjamin H Savitzky, Andrew M Minor, and MC Scott. Automated crystal orientation mapping in py4dstem using sparse correlation matching. Microscopy and Microanalysis, 28:390-403, Oct 2022. URL: https://doi.org/10.48550/arxiv.2111.00171, doi:10.48550/arxiv.2111.00171. This article has 115 citations and is from a peer-reviewed journal." + } + ], + "call_sXDuxIhQcjT0pGraCjqsLz2r": [ + { + "embedding": null, + "docname": "wanni2024machinelearningenhanced", + "dockey": "bf3e624ed86e8966", + "citation": "J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "wanni2024machinelearningenhanced", + "bibtex": "@article{wanni2024machinelearningenhanced,\n author = \"Wanni, J. and Bronkhorst, C. A. and Thoma, D. J.\",\n title = \"Machine learning enhanced analysis of EBSD data for texture representation\",\n year = \"2024\",\n journal = \"npj Computational Materials\",\n volume = \"10\",\n pages = \"1-11\",\n month = \"Jun\",\n doi = \"10.1038/s41524-024-01324-4\",\n url = \"https://doi.org/10.1038/s41524-024-01324-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n", + "authors": [ + "J. Wanni", + "C. A. Bronkhorst", + "D. J. Thoma" + ], + "publication_date": "2024-06-26T00:00:00Z", + "year": 2024, + "volume": "10", + "issue": "1", + "issn": "2057-3960", + "pages": "1-11", + "journal": "npj Computational Materials", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41524-024-01324-4", + "title": "Machine learning enhanced analysis of EBSD data for texture representation", + "citation_count": 14, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41524-024-01324-4", + "doi_url": "https://doi.org/10.1038/s41524-024-01324-4", + "doc_id": "bf3e624ed86e8966", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41524-024-01324-4.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Wanni, C. A. Bronkhorst, and D. J. Thoma. Machine learning enhanced analysis of ebsd data for texture representation. npj Computational Materials, 10:1-11, Jun 2024. URL: https://doi.org/10.1038/s41524-024-01324-4, doi:10.1038/s41524-024-01324-4. This article has 14 citations and is from a peer-reviewed journal." + } + ], + "call_I26KQDTI4oHtsQAWeLYGK9RU": [], + "call_5NDupt5Es4nrgT9SSGt0veGH": [ + { + "embedding": null, + "docname": "ostormujof2022deeplearningfor", + "dockey": "ddc178136b11c42d", + "citation": "T Martinez Ostormujof, Rrp Purushottam Raj Purohit, S Breumier, Nathalie Gey, M Salib, and L Germain. Deep learning for automated phase segmentation in ebsd maps. a case study in dual phase steel microstructures. Materials Characterization, Nov 2022. URL: https://doi.org/10.48550/arxiv.2112.03072, doi:10.48550/arxiv.2112.03072.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "ostormujof2022deeplearningfor", + "bibtex": "@article{ostormujof2022deeplearningfor,\n author = \"Ostormujof, T Martinez and Purohit, Rrp Purushottam Raj and Breumier, S and Gey, Nathalie and Salib, M and Germain, L\",\n title = \"Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures\",\n year = \"2022\",\n journal = \"Materials Characterization\",\n month = \"Nov\",\n doi = \"10.48550/arxiv.2112.03072\",\n url = \"https://doi.org/10.48550/arxiv.2112.03072\",\n publisher = \"arXiv\"\n}\n", + "authors": [ + "T Martinez Ostormujof", + "Rrp Purushottam Raj Purohit", + "S Breumier", + "Nathalie Gey", + "M Salib", + "L Germain" + ], + "publication_date": "2021-11-26T00:00:00Z", + "year": 2022, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Materials Characterization", + "publisher": "arXiv", + "url": "https://arxiv.org/abs/2112.03072", + "title": "Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures", + "citation_count": 65, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.48550/arxiv.2112.03072", + "doi_url": "https://doi.org/10.48550/arxiv.2112.03072", + "doc_id": "ddc178136b11c42d", + "file_location": null, + "license": null, + "pdf_url": "https://www.sciencedirect.com/science/article/am/pii/S1044580321007609", + "other": { + "client_source": [ + "paper_search", + "datacite", + "semantic_scholar", + "datacite", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "T Martinez Ostormujof, Rrp Purushottam Raj Purohit, S Breumier, Nathalie Gey, M Salib, and L Germain. Deep learning for automated phase segmentation in ebsd maps. a case study in dual phase steel microstructures. Materials Characterization, Nov 2022. URL: https://doi.org/10.48550/arxiv.2112.03072, doi:10.48550/arxiv.2112.03072. This article has 65 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "taufique2025generalizableimagesegmentation", + "dockey": "c03e3381d9760a6e", + "citation": "M. F. N. Taufique, Julia Nguyen, Julian D. Escobar, Mayur Pole, David Garcia, Tianhao Wang, Hrishikesh Das, Kathy Nwe, David Brown, Kenneth Ross, Donald R. Todd, Erin Barker, Eric Smith, Keerti S. Kappagantula, and Jenna A. Bilbrey. Generalizable image segmentation for microstructure characterization through integrated sem and ebsd analysis. npj Computational Materials, Oct 2025. URL: https://doi.org/10.1038/s41524-025-01801-4, doi:10.1038/s41524-025-01801-4.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "taufique2025generalizableimagesegmentation", + "bibtex": "@article{taufique2025generalizableimagesegmentation,\n author = \"Taufique, M. F. N. and Nguyen, Julia and Escobar, Julian D. and Pole, Mayur and Garcia, David and Wang, Tianhao and Das, Hrishikesh and Nwe, Kathy and Brown, David and Ross, Kenneth and Todd, Donald R. and Barker, Erin and Smith, Eric and Kappagantula, Keerti S. and Bilbrey, Jenna A.\",\n title = \"Generalizable image segmentation for microstructure characterization through integrated SEM and EBSD analysis\",\n year = \"2025\",\n journal = \"npj Computational Materials\",\n volume = \"11\",\n month = \"Oct\",\n doi = \"10.1038/s41524-025-01801-4\",\n url = \"https://doi.org/10.1038/s41524-025-01801-4\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2057-3960\"\n}\n", + "authors": [ + "M. F. N. Taufique", + "Julia Nguyen", + "Julian D. Escobar", + "Mayur Pole", + "David Garcia", + "Tianhao Wang", + "Hrishikesh Das", + "Kathy Nwe", + "David Brown", + "Kenneth Ross", + "Donald R. Todd", + "Erin Barker", + "Eric Smith", + "Keerti S. Kappagantula", + "Jenna A. Bilbrey" + ], + "publication_date": "2025-10-29T00:00:00Z", + "year": 2025, + "volume": "11", + "issue": "1", + "issn": "2057-3960", + "pages": null, + "journal": "npj Computational Materials", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1038/s41524-025-01801-4", + "title": "Generalizable image segmentation for microstructure characterization through integrated SEM and EBSD analysis", + "citation_count": 6, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1038/s41524-025-01801-4", + "doi_url": "https://doi.org/10.1038/s41524-025-01801-4", + "doc_id": "c03e3381d9760a6e", + "file_location": null, + "license": null, + "pdf_url": "https://www.nature.com/articles/s41524-025-01801-4.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "M. F. N. Taufique, Julia Nguyen, Julian D. Escobar, Mayur Pole, David Garcia, Tianhao Wang, Hrishikesh Das, Kathy Nwe, David Brown, Kenneth Ross, Donald R. Todd, Erin Barker, Eric Smith, Keerti S. Kappagantula, and Jenna A. Bilbrey. Generalizable image segmentation for microstructure characterization through integrated sem and ebsd analysis. npj Computational Materials, Oct 2025. URL: https://doi.org/10.1038/s41524-025-01801-4, doi:10.1038/s41524-025-01801-4. This article has 6 citations and is from a peer-reviewed journal." + } + ], + "call_TcPaoKE5klpaxx5d8PlBUDrk": [ + { + "embedding": null, + "docname": "ball2023registrationbetweendct", + "dockey": "0a036b5088d09190", + "citation": "J. Ball, J. Oddershede, C. Davis, C. Slater, M. Said, H. Vashishtha, \u0160. Michalik, David M. Collins School of Metallurgy, Materials, U. Birmingham, United Kingdom., Diamond Light Source Ltd., Xnovo Technology ApS, Denmark, Wmg, and U. Warwick. Registration between dct and ebsd datasets for multiphase microstructures. Materials Characterization, 204:113228, Oct 2023. URL: https://doi.org/10.1016/j.matchar.2023.113228, doi:10.1016/j.matchar.2023.113228.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "ball2023registrationbetweendct", + "bibtex": "@article{ball2023registrationbetweendct,\n author = \"Ball, J. and Oddershede, J. and Davis, C. and Slater, C. and Said, M. and Vashishtha, H. and Michalik, \u0160. and of Metallurgy, David M. Collins School and Materials and Birmingham, U. and Kingdom., United and Ltd., Diamond Light Source and ApS, Xnovo Technology and Denmark and Wmg and Warwick, U.\",\n title = \"Registration between DCT and EBSD datasets for multiphase microstructures\",\n year = \"2023\",\n journal = \"Materials Characterization\",\n volume = \"204\",\n pages = \"113228\",\n month = \"Oct\",\n doi = \"10.1016/j.matchar.2023.113228\",\n url = \"https://doi.org/10.1016/j.matchar.2023.113228\",\n publisher = \"Elsevier BV\",\n issn = \"1044-5803\"\n}\n", + "authors": [ + "J. Ball", + "J. Oddershede", + "C. Davis", + "C. Slater", + "M. Said", + "H. Vashishtha", + "\u0160. Michalik", + "David M. Collins School of Metallurgy", + "Materials", + "U. Birmingham", + "United Kingdom.", + "Diamond Light Source Ltd.", + "Xnovo Technology ApS", + "Denmark", + "Wmg", + "U. Warwick" + ], + "publication_date": "2023-10-01T00:00:00Z", + "year": 2023, + "volume": "204", + "issue": null, + "issn": "1044-5803", + "pages": "113228", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2023.113228", + "title": "Registration between DCT and EBSD datasets for multiphase microstructures", + "citation_count": 8, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2023.113228", + "doi_url": "https://doi.org/10.1016/j.matchar.2023.113228", + "doc_id": "0a036b5088d09190", + "file_location": null, + "license": null, + "pdf_url": "https://ars.els-cdn.com/content/image/1-s2.0-S1044580323005879-ga1_lrg.jpg", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "J. Ball, J. Oddershede, C. Davis, C. Slater, M. Said, H. Vashishtha, \u0160. Michalik, David M. Collins School of Metallurgy, Materials, U. Birmingham, United Kingdom., Diamond Light Source Ltd., Xnovo Technology ApS, Denmark, Wmg, and U. Warwick. Registration between dct and ebsd datasets for multiphase microstructures. Materials Characterization, 204:113228, Oct 2023. URL: https://doi.org/10.1016/j.matchar.2023.113228, doi:10.1016/j.matchar.2023.113228. This article has 8 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "polonsky2026pytribeamopensourcesoftware", + "dockey": "4deb79e3fdff92cc", + "citation": "Andrew T. Polonsky, James D. Lamb, Chad B. Hovey, Haydn Schroader, McLean P. Echlin, and Tresa M. Pollock. Pytribeam: open-source software for enhanced 3d data collection in tribeam microscopes. Integrating Materials and Manufacturing Innovation, 15:1-17, Feb 2026. URL: https://doi.org/10.1007/s40192-026-00449-2, doi:10.1007/s40192-026-00449-2.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2026pytribeamopensourcesoftware", + "bibtex": "@article{polonsky2026pytribeamopensourcesoftware,\n author = \"Polonsky, Andrew T. and Lamb, James D. and Hovey, Chad B. and Schroader, Haydn and Echlin, McLean P. and Pollock, Tresa M.\",\n title = \"pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes\",\n year = \"2026\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"15\",\n pages = \"1-17\",\n month = \"Feb\",\n doi = \"10.1007/s40192-026-00449-2\",\n url = \"https://doi.org/10.1007/s40192-026-00449-2\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "James D. Lamb", + "Chad B. Hovey", + "Haydn Schroader", + "McLean P. Echlin", + "Tresa M. Pollock" + ], + "publication_date": "2026-02-27T00:00:00Z", + "year": 2026, + "volume": "15", + "issue": "1", + "issn": "2193-9764", + "pages": "1-17", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-026-00449-2", + "title": "pyTriBeam: Open-Source Software for Enhanced 3D Data Collection in TriBeam Microscopes", + "citation_count": 0, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-026-00449-2", + "doi_url": "https://doi.org/10.1007/s40192-026-00449-2", + "doc_id": "4deb79e3fdff92cc", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-026-00449-2.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, James D. Lamb, Chad B. Hovey, Haydn Schroader, McLean P. Echlin, and Tresa M. Pollock. Pytribeam: open-source software for enhanced 3d data collection in tribeam microscopes. Integrating Materials and Manufacturing Innovation, 15:1-17, Feb 2026. URL: https://doi.org/10.1007/s40192-026-00449-2, doi:10.1007/s40192-026-00449-2. This article has 0 citations and is from a peer-reviewed journal." + } + ], + "call_YCK1KFOsT4kHD9IBqs5QYRHi": [], + "call_YAEYGtulkKwxKg8CZdFRkuKJ": [], + "call_6EDjDZUx9EcloPKavvbsh1xb": [], + "call_bd44zev0ajEnct80hGQyaouJ": [], + "call_Z5I4Dszqv8IduAWdkGxzojFr": [ + { + "embedding": null, + "docname": "chapman2021afrladditivemanufacturing", + "dockey": "729959244c4b906f", + "citation": "Michael G. Chapman, Megna N. Shah, Sean P. Donegan, J. Michael Scott, Paul A. Shade, David Menasche, and Michael D. Uchic. Afrl additive manufacturing modeling series: challenge 4, 3d reconstruction of an in625 high-energy diffraction microscopy sample using multi-modal serial sectioning. Integrating Materials and Manufacturing Innovation, 10:129-141, Jun 2021. URL: https://doi.org/10.1007/s40192-021-00212-9, doi:10.1007/s40192-021-00212-9.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "chapman2021afrladditivemanufacturing", + "bibtex": "@article{chapman2021afrladditivemanufacturing,\n author = \"Chapman, Michael G. and Shah, Megna N. and Donegan, Sean P. and Scott, J. Michael and Shade, Paul A. and Menasche, David and Uchic, Michael D.\",\n title = \"AFRL Additive Manufacturing Modeling Series: Challenge 4, 3D Reconstruction of an IN625 High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning\",\n year = \"2021\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"10\",\n pages = \"129-141\",\n month = \"Jun\",\n doi = \"10.1007/s40192-021-00212-9\",\n url = \"https://doi.org/10.1007/s40192-021-00212-9\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Michael G. Chapman", + "Megna N. Shah", + "Sean P. Donegan", + "J. Michael Scott", + "Paul A. Shade", + "David Menasche", + "Michael D. Uchic" + ], + "publication_date": "2021-06-01T00:00:00Z", + "year": 2021, + "volume": "10", + "issue": "2", + "issn": "2193-9764", + "pages": "129-141", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-021-00212-9", + "title": "AFRL Additive Manufacturing Modeling Series: Challenge 4, 3D Reconstruction of an IN625 High-Energy Diffraction Microscopy Sample Using Multi-modal Serial Sectioning", + "citation_count": 38, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-021-00212-9", + "doi_url": "https://doi.org/10.1007/s40192-021-00212-9", + "doc_id": "729959244c4b906f", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-021-00212-9.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Michael G. Chapman, Megna N. Shah, Sean P. Donegan, J. Michael Scott, Paul A. Shade, David Menasche, and Michael D. Uchic. Afrl additive manufacturing modeling series: challenge 4, 3d reconstruction of an in625 high-energy diffraction microscopy sample using multi-modal serial sectioning. Integrating Materials and Manufacturing Innovation, 10:129-141, Jun 2021. URL: https://doi.org/10.1007/s40192-021-00212-9, doi:10.1007/s40192-021-00212-9. This article has 38 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "clarke2024frameworkforthe", + "dockey": "d0af44c0fc36011d", + "citation": "Kenneth M. Clarke, John Wertz, and Michael Groeber. Framework for the generation of 3d fiber composite structures from 2d observations. Integrating Materials and Manufacturing Innovation, 13:569-584, May 2024. URL: https://doi.org/10.1007/s40192-024-00352-8, doi:10.1007/s40192-024-00352-8.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "clarke2024frameworkforthe", + "bibtex": "@article{clarke2024frameworkforthe,\n author = \"Clarke, Kenneth M. and Wertz, John and Groeber, Michael\",\n title = \"Framework for the Generation of 3D Fiber Composite Structures from 2D Observations\",\n year = \"2024\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"13\",\n pages = \"569-584\",\n month = \"May\",\n doi = \"10.1007/s40192-024-00352-8\",\n url = \"https://doi.org/10.1007/s40192-024-00352-8\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Kenneth M. Clarke", + "John Wertz", + "Michael Groeber" + ], + "publication_date": "2024-05-16T00:00:00Z", + "year": 2024, + "volume": "13", + "issue": "2", + "issn": "2193-9764", + "pages": "569-584", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-024-00352-8", + "title": "Framework for the Generation of 3D Fiber Composite Structures from 2D Observations", + "citation_count": 1, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-024-00352-8", + "doi_url": "https://doi.org/10.1007/s40192-024-00352-8", + "doc_id": "d0af44c0fc36011d", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-024-00352-8.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Kenneth M. Clarke, John Wertz, and Michael Groeber. Framework for the generation of 3d fiber composite structures from 2d observations. Integrating Materials and Manufacturing Innovation, 13:569-584, May 2024. URL: https://doi.org/10.1007/s40192-024-00352-8, doi:10.1007/s40192-024-00352-8. This article has 1 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "zukic2021itkmontageasoftware", + "dockey": "a598bbb0f2f4b9e2", + "citation": "D\u017eenan Zuki\u0107, Michael Jackson, Dennis Dimiduk, Sean Donegan, Michael Groeber, and Matthew McCormick. Itkmontage: a software module for image stitching. Integrating Materials and Manufacturing Innovation, 10:115-124, Mar 2021. URL: https://doi.org/10.1007/s40192-021-00202-x, doi:10.1007/s40192-021-00202-x.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "zukic2021itkmontageasoftware", + "bibtex": "@article{zukic2021itkmontageasoftware,\n author = \"Zuki\u0107, D\u017eenan and Jackson, Michael and Dimiduk, Dennis and Donegan, Sean and Groeber, Michael and McCormick, Matthew\",\n title = \"ITKMontage: A Software Module for Image Stitching\",\n year = \"2021\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"10\",\n pages = \"115-124\",\n month = \"Mar\",\n doi = \"10.1007/s40192-021-00202-x\",\n url = \"https://doi.org/10.1007/s40192-021-00202-x\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "D\u017eenan Zuki\u0107", + "Michael Jackson", + "Dennis Dimiduk", + "Sean Donegan", + "Michael Groeber", + "Matthew McCormick" + ], + "publication_date": "2021-03-01T00:00:00Z", + "year": 2021, + "volume": "10", + "issue": "1", + "issn": "2193-9764", + "pages": "115-124", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-021-00202-x", + "title": "ITKMontage: A Software Module for Image Stitching", + "citation_count": 25, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-021-00202-x", + "doi_url": "https://doi.org/10.1007/s40192-021-00202-x", + "doc_id": "a598bbb0f2f4b9e2", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-021-00202-x.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "D\u017eenan Zuki\u0107, Michael Jackson, Dennis Dimiduk, Sean Donegan, Michael Groeber, and Matthew McCormick. Itkmontage: a software module for image stitching. Integrating Materials and Manufacturing Innovation, 10:115-124, Mar 2021. URL: https://doi.org/10.1007/s40192-021-00202-x, doi:10.1007/s40192-021-00202-x. This article has 25 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform", + "dockey": "c19862576a1c6ce9", + "citation": "L. Koschmieder, S. Hojda, M. Apel, R. Altenfeld, Y. Bami, C. Haase, M. Lin, A. Vuppala, G. Hirt, and G.J. Schmitz. Aixvipmap\u00ae\u2014an operational platform for microstructure modeling workflows. Integrating Materials and Manufacturing Innovation, 8:122-143, Jun 2019. URL: https://doi.org/10.1007/s40192-019-00138-3, doi:10.1007/s40192-019-00138-3.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform", + "bibtex": "@article{koschmieder2019aixvipmap\u00ae\u2014anoperationalplatform,\n author = \"Koschmieder, L. and Hojda, S. and Apel, M. and Altenfeld, R. and Bami, Y. and Haase, C. and Lin, M. and Vuppala, A. and Hirt, G. and Schmitz, G.J.\",\n title = \"AixViPMaP\u00ae\u2014an Operational Platform for Microstructure Modeling Workflows\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"122-143\",\n month = \"Jun\",\n doi = \"10.1007/s40192-019-00138-3\",\n url = \"https://doi.org/10.1007/s40192-019-00138-3\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"2\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "L. Koschmieder", + "S. Hojda", + "M. Apel", + "R. Altenfeld", + "Y. Bami", + "C. Haase", + "M. Lin", + "A. Vuppala", + "G. Hirt", + "G.J. Schmitz" + ], + "publication_date": "2019-06-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "2", + "issn": "2193-9764", + "pages": "122-143", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://doi.org/10.1007/s40192-019-00138-3", + "title": "AixViPMaP\u00ae\u2014an Operational Platform for Microstructure Modeling Workflows", + "citation_count": 14, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00138-3", + "doi_url": "https://doi.org/10.1007/s40192-019-00138-3", + "doc_id": "c19862576a1c6ce9", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00138-3.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "L. Koschmieder, S. Hojda, M. Apel, R. Altenfeld, Y. Bami, C. Haase, M. Lin, A. Vuppala, G. Hirt, and G.J. Schmitz. Aixvipmap\u00ae\u2014an operational platform for microstructure modeling workflows. Integrating Materials and Manufacturing Innovation, 8:122-143, Jun 2019. URL: https://doi.org/10.1007/s40192-019-00138-3, doi:10.1007/s40192-019-00138-3. This article has 14 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "polonsky2019threedimensionalanalysisand", + "dockey": "39e5f645a50ad2d5", + "citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "polonsky2019threedimensionalanalysisand", + "bibtex": "@article{polonsky2019threedimensionalanalysisand,\n author = \"Polonsky, Andrew T. and Lang, Christian A. and Kvilekval, Kristian G. and Latypov, Marat I. and Echlin, McLean P. and Manjunath, B. S. and Pollock, Tresa M.\",\n title = \"Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure\",\n year = \"2019\",\n journal = \"Integrating Materials and Manufacturing Innovation\",\n volume = \"8\",\n pages = \"37-51\",\n month = \"Mar\",\n doi = \"10.1007/s40192-019-00126-7\",\n url = \"https://doi.org/10.1007/s40192-019-00126-7\",\n publisher = \"Springer Science and Business Media LLC\",\n issue = \"1\",\n issn = \"2193-9764\"\n}\n", + "authors": [ + "Andrew T. Polonsky", + "Christian A. Lang", + "Kristian G. Kvilekval", + "Marat I. Latypov", + "McLean P. Echlin", + "B. S. Manjunath", + "Tresa M. Pollock" + ], + "publication_date": "2019-03-01T00:00:00Z", + "year": 2019, + "volume": "8", + "issue": "1", + "issn": "2193-9764", + "pages": "37-51", + "journal": "Integrating Materials and Manufacturing Innovation", + "publisher": "Springer Science and Business Media LLC", + "url": "https://link.springer.com/content/pdf/10.1007%2Fs40192-019-00126-7.pdf", + "title": "Three-dimensional Analysis and Reconstruction of Additively Manufactured Materials in the Cloud-Based BisQue Infrastructure", + "citation_count": 31, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1007/s40192-019-00126-7", + "doi_url": "https://doi.org/10.1007/s40192-019-00126-7", + "doc_id": "39e5f645a50ad2d5", + "file_location": null, + "license": null, + "pdf_url": "https://link.springer.com/content/pdf/10.1007/s40192-019-00126-7.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "semantic_scholar", + "crossref", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated", + "self_generated", + [ + "self_generated" + ], + [ + "self_generated" + ], + "self_generated" + ] + }, + "formatted_citation": "Andrew T. Polonsky, Christian A. Lang, Kristian G. Kvilekval, Marat I. Latypov, McLean P. Echlin, B. S. Manjunath, and Tresa M. Pollock. Three-dimensional analysis and reconstruction of additively manufactured materials in the cloud-based bisque infrastructure. Integrating Materials and Manufacturing Innovation, 8:37-51, Mar 2019. URL: https://doi.org/10.1007/s40192-019-00126-7, doi:10.1007/s40192-019-00126-7. This article has 31 citations and is from a peer-reviewed journal." + } + ], + "call_5hiGgVfNiBeJ5xEkpggCPT0O": [ + { + "embedding": null, + "docname": "savage2021anautomatedprocedure", + "dockey": "22c76dc107283440", + "citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "savage2021anautomatedprocedure", + "bibtex": "@article{savage2021anautomatedprocedure,\n author = \"Savage, Daniel J. and McCabe, Rodney J. and Knezevic, Marko\",\n title = \"An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures\",\n volume = \"171\",\n issn = \"1044-5803\",\n url = \"https://doi.org/10.1016/j.matchar.2020.110808\",\n doi = \"10.1016/j.matchar.2020.110808\",\n journal = \"Materials Characterization\",\n publisher = \"Elsevier BV\",\n year = \"2021\",\n month = \"Jan\",\n pages = \"110808\"\n}\n", + "authors": [ + "Daniel J. Savage", + "Rodney J. McCabe", + "Marko Knezevic" + ], + "publication_date": "2021-01-01T00:00:00Z", + "year": 2021, + "volume": "171", + "issue": null, + "issn": "1044-5803", + "pages": "110808", + "journal": "Materials Characterization", + "publisher": "Elsevier BV", + "url": "https://doi.org/10.1016/j.matchar.2020.110808", + "title": "An automated procedure built on MTEX for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures", + "citation_count": 20, + "bibtex_type": "misc", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1016/j.matchar.2020.110808", + "doi_url": "https://doi.org/10.1016/j.matchar.2020.110808", + "doc_id": "22c76dc107283440", + "file_location": null, + "license": null, + "pdf_url": "http://manuscript.elsevier.com/S1044580320322798/pdf/S1044580320322798.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + "semantic_scholar", + "self_generated", + "semantic_scholar", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "Daniel J. Savage, Rodney J. McCabe, and Marko Knezevic. An automated procedure built on mtex for reconstructing deformation twin hierarchies from electron backscattered diffraction datasets of heavily twinned microstructures. Materials Characterization, 171:110808, Jan 2021. URL: https://doi.org/10.1016/j.matchar.2020.110808, doi:10.1016/j.matchar.2020.110808. This article has 20 citations and is from a peer-reviewed journal." + }, + { + "embedding": null, + "docname": "chateigner2019quantitativetextureanalysis", + "dockey": "0558c5eac2b5ce39", + "citation": "D. Chateigner, L. Lutterotti, and M. Morales. Quantitative texture analysis and combined analysis, pages 555-580. International Union of Crystallography, Dec 2019. URL: https://doi.org/10.1107/97809553602060000968, doi:10.1107/97809553602060000968.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "chateigner2019quantitativetextureanalysis", + "bibtex": "@inbook{chateigner2019quantitativetextureanalysis,\n author = \"Chateigner, D. and Lutterotti, L. and Morales, M.\",\n title = \"Quantitative texture analysis and combined analysis\",\n ISBN = \"9781118416280\",\n url = \"https://doi.org/10.1107/97809553602060000968\",\n doi = \"10.1107/97809553602060000968\",\n booktitle = \"International Tables for Crystallography\",\n publisher = \"International Union of Crystallography\",\n year = \"2019\",\n month = \"Dec\",\n pages = \"555-580\",\n journal = \"International Tables for Crystallography\"\n}\n", + "authors": [ + "D. Chateigner", + "L. Lutterotti", + "M. Morales" + ], + "publication_date": "2007-12-12T00:00:00Z", + "year": 2019, + "volume": null, + "issue": null, + "issn": null, + "pages": "555-580", + "journal": "International Tables for Crystallography", + "publisher": "International Union of Crystallography", + "url": "https://doi.org/10.1107/97809553602060000968", + "title": "Quantitative texture analysis and combined analysis", + "citation_count": 9, + "bibtex_type": "inbook", + "source_quality": -1, + "is_retracted": null, + "doi": "10.1107/97809553602060000968", + "doi_url": "https://doi.org/10.1107/97809553602060000968", + "doc_id": "0558c5eac2b5ce39", + "file_location": null, + "license": null, + "pdf_url": "https://chateigner.ensicaen.fr/pdf/Chateigner_ITC_Vol%20H%20Chapter%205_3_555_2019_Quantitative%20texture%20analysis%20and%20combined%20analysis.pdf", + "other": { + "client_source": [ + "paper_search", + "crossref", + "crossref" + ], + "bibtex_source": [ + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ], + "self_generated", + [ + "self_generated" + ], + [ + "crossref" + ] + ] + }, + "formatted_citation": "D. Chateigner, L. Lutterotti, and M. Morales. Quantitative texture analysis and combined analysis, pages 555-580. International Union of Crystallography, Dec 2019. URL: https://doi.org/10.1107/97809553602060000968, doi:10.1107/97809553602060000968. This article has 9 citations." + }, + { + "embedding": null, + "docname": "einhorn2017asurveyand", + "dockey": "3f5fd0caf32ed21e", + "citation": "John Einhorn. A survey and comparison of methods for quantitative textureanalysis. ArXiv, Jul 2017. URL: https://doi.org/10.18130/v30r4d, doi:10.18130/v30r4d.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "einhorn2017asurveyand", + "bibtex": "@article{einhorn2017asurveyand,\n author = \"Einhorn, John\",\n title = \"A Survey and Comparison of Methods for Quantitative TextureAnalysis\",\n year = \"2017\",\n journal = \"ArXiv\",\n doi = \"10.18130/v30r4d\",\n url = \"https://doi.org/10.18130/v30r4d\",\n month = \"Jul\",\n publisher = \"University of Virginia\"\n}\n", + "authors": [ + "John Einhorn" + ], + "publication_date": "2017-07-31T00:00:00Z", + "year": 2017, + "volume": "", + "issue": null, + "issn": null, + "pages": null, + "journal": "ArXiv", + "publisher": "University of Virginia", + "url": "https://libraetd.lib.virginia.edu/downloads/1g05fb752?filename=1_Einhorn_John_2017_MS.pdf", + "title": "A Survey and Comparison of Methods for Quantitative TextureAnalysis", + "citation_count": 4, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.18130/v30r4d", + "doi_url": "https://doi.org/10.18130/v30r4d", + "doc_id": "3f5fd0caf32ed21e", + "file_location": null, + "license": null, + "pdf_url": "https://scholar.archive.org/work/vjjprd3k2fgrplzwv54ctyv3re/access/wayback/https://libraetd.lib.virginia.edu/downloads/1g05fb752?filename=1_Einhorn_John_2017_MS.pdf", + "other": { + "client_source": [ + "paper_search", + "datacite", + "datacite", + "semantic_scholar", + "datacite", + "datacite", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "John Einhorn. A survey and comparison of methods for quantitative textureanalysis. ArXiv, Jul 2017. URL: https://doi.org/10.18130/v30r4d, doi:10.18130/v30r4d. This article has 4 citations." + } + ], + "call_t3X9GkiKh5YtIwvvFGFtqFju": [], + "call_2WFpaGkkorc7QNUjgoGtFJOi": [ + { + "embedding": null, + "docname": "mukherjee2022aroadmapfor", + "dockey": "332e927f945f4400", + "citation": "Debangshu Mukherjee, Kevin M. Roccapriore, Anees Al-Najjar, Ayana Ghosh, Jacob D. Hinkle, Andrew R. Lupini, Rama K. Vasudevan, Sergei V. Kalinin, Olga S. Ovchinnikova, Maxim A. Ziatdinov, and Nageswara S. Rao. A roadmap for edge computing enabled automated multidimensional transmission electron microscopy. Microscopy Today, 30:10-19, Oct 2022. URL: https://doi.org/10.48550/arxiv.2210.02538, doi:10.48550/arxiv.2210.02538.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "mukherjee2022aroadmapfor", + "bibtex": "@article{mukherjee2022aroadmapfor,\n author = \"Mukherjee, Debangshu and Roccapriore, Kevin M. and Al-Najjar, Anees and Ghosh, Ayana and Hinkle, Jacob D. and Lupini, Andrew R. and Vasudevan, Rama K. and Kalinin, Sergei V. and Ovchinnikova, Olga S. and Ziatdinov, Maxim A. and Rao, Nageswara S.\",\n title = \"A Roadmap for Edge Computing Enabled Automated Multidimensional Transmission Electron Microscopy\",\n year = \"2022\",\n journal = \"Microscopy Today\",\n volume = \"30\",\n pages = \"10-19\",\n month = \"Oct\",\n doi = \"10.48550/arxiv.2210.02538\",\n url = \"https://doi.org/10.48550/arxiv.2210.02538\",\n publisher = \"arXiv\"\n}\n", + "authors": [ + "Debangshu Mukherjee", + "Kevin M. Roccapriore", + "Anees Al-Najjar", + "Ayana Ghosh", + "Jacob D. Hinkle", + "Andrew R. Lupini", + "Rama K. Vasudevan", + "Sergei V. Kalinin", + "Olga S. Ovchinnikova", + "Maxim A. Ziatdinov", + "Nageswara S. Rao" + ], + "publication_date": "2022-10-05T00:00:00Z", + "year": 2022, + "volume": "30", + "issue": null, + "issn": null, + "pages": "10-19", + "journal": "Microscopy Today", + "publisher": "arXiv", + "url": "https://arxiv.org/abs/2210.02538", + "title": "A Roadmap for Edge Computing Enabled Automated Multidimensional Transmission Electron Microscopy", + "citation_count": 23, + "bibtex_type": "article", + "source_quality": -1, + "is_retracted": null, + "doi": "10.48550/arxiv.2210.02538", + "doi_url": "https://doi.org/10.48550/arxiv.2210.02538", + "doc_id": "332e927f945f4400", + "file_location": null, + "license": null, + "pdf_url": "https://arxiv.org/pdf/2210.02538", + "other": { + "client_source": [ + "paper_search", + "datacite", + "semantic_scholar", + "datacite", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Debangshu Mukherjee, Kevin M. Roccapriore, Anees Al-Najjar, Ayana Ghosh, Jacob D. Hinkle, Andrew R. Lupini, Rama K. Vasudevan, Sergei V. Kalinin, Olga S. Ovchinnikova, Maxim A. Ziatdinov, and Nageswara S. Rao. A roadmap for edge computing enabled automated multidimensional transmission electron microscopy. Microscopy Today, 30:10-19, Oct 2022. URL: https://doi.org/10.48550/arxiv.2210.02538, doi:10.48550/arxiv.2210.02538. This article has 23 citations." + }, + { + "embedding": null, + "docname": "botifoll2022machinelearningin", + "dockey": "c976a9b5ef1897d8", + "citation": "Marc Botifoll, Ivan Pinto-Huguet, and J. Arbiol. Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook. Nanoscale horizons, Oct 2022. URL: https://doi.org/10.1039/d2nh00377e, doi:10.1039/d2nh00377e.", + "content_hash": null, + "fields_to_overwrite_from_metadata": [ + "citation", + "content_hash", + "doc_id", + "dockey", + "docname", + "key" + ], + "key": "botifoll2022machinelearningin", + "bibtex": "@article{botifoll2022machinelearningin,\n author = \"Botifoll, Marc and Pinto-Huguet, Ivan and Arbiol, J.\",\n title = \"Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook.\",\n year = \"2022\",\n journal = \"Nanoscale horizons\",\n month = \"Oct\",\n doi = \"10.1039/d2nh00377e\",\n url = \"https://doi.org/10.1039/d2nh00377e\"\n}\n", + "authors": [ + "Marc Botifoll", + "Ivan Pinto-Huguet", + "J. Arbiol" + ], + "publication_date": "2022-10-14T00:00:00Z", + "year": 2022, + "volume": null, + "issue": null, + "issn": null, + "pages": null, + "journal": "Nanoscale horizons", + "publisher": null, + "url": "https://digital.csic.es/bitstream/10261/293145/1/Machine_learning_Botifoll.pdf", + "title": "Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook.", + "citation_count": 159, + "bibtex_type": "article", + "source_quality": 1, + "is_retracted": null, + "doi": "10.1039/d2nh00377e", + "doi_url": "https://doi.org/10.1039/d2nh00377e", + "doc_id": "c976a9b5ef1897d8", + "file_location": null, + "license": null, + "pdf_url": "https://digital.csic.es/bitstream/10261/293145/1/Machine_learning_Botifoll.pdf", + "other": { + "client_source": [ + "paper_search", + "semantic_scholar", + "semantic_scholar", + "semantic_scholar", + "semantic_scholar" + ], + "bibtex_source": [ + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated", + "self_generated" + ] + }, + "formatted_citation": "Marc Botifoll, Ivan Pinto-Huguet, and J. Arbiol. Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook. Nanoscale horizons, Oct 2022. URL: https://doi.org/10.1039/d2nh00377e, doi:10.1039/d2nh00377e. This article has 159 citations and is from a peer-reviewed journal." + } + ], + "call_xqX7V5hmrKT4tjzjK7lLSxlE": [], + "call_QhQG5TaOdTjPfKLBDrG4HStR": [], + "call_at4kevxNAZzrz4Vmrs3X4MTb": [], + "call_1oiOo0bOuigk4ZUPXbdj7xZ2": [], + "call_eVCSSI1d7fc6pWTEoIMUSTF0": [], + "call_UpEp4rBNUwkVd82OjFFGqJTn": [], + "call_P8jQvXEYp1OGYeb95uE317Pf": [], + "call_QJeM4kTCkLbZRr827PUgMWr1": [], + "call_ibuikCX928XmGVA9Di78lliY": [], + "call_FRljLsn8HFS1KSTBdRAWzMFk": [], + "call_7bcoFGw2YuGzai5ip8f8B7P4": [], + "call_0Ow8AsXKD75JrGG2nzDUQk6Q": [], + "call_8LdlgnIRhAHn08c2v67pYdEC": [] + }, + "used_contexts": [ + "pqac-00000009", + "pqac-00000005", + "pqac-00000006", + "pqac-00000007", + "pqac-00000008", + "pqac-00000001", + "pqac-00000013", + "pqac-00000010", + "pqac-00000002", + "pqac-00000003", + "pqac-00000012", + "pqac-00000004", + "pqac-00000011", + "pqac-00000000" + ] + }, + "bibtex": null, + "status": "success", + "timing_info": null, + "duration": 0.0, + "stats": null, + "usage": {} + }, + "cost": 0.0, + "permission": "private", + "query_group": null + }, + "info": { + "query": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB." + } + }, + "supplemental_data": null, + "created_at": "2026-07-29T03:41:37.609612Z" + }, + "metadata": null, + "deployment_config": null, + "failure_reason": null, + "task": "What are the open-source, Python-native software packages for electron backscatter diffraction (EBSD) post-processing that serve as alternatives to the MATLAB-based MTEX toolbox? Focus on capabilities for grain reconstruction/segmentation from orientation maps, crystallographic texture analysis (ODF, pole figures, inverse pole figures), and misorientation-derived metrics such as kernel average misorientation (KAM) and geometrically necessary dislocation (GND) density. Discuss packages such as kikuchipy, orix, DefDAP, DREAM.3D/DREAM3D-NX (simplnx), pyxem, and any others, including their maturity, peer-reviewed references, and how their grain-reconstruction and texture-analysis capabilities compare with MTEX. Context: an automated Python pipeline (indexing via kikuchipy/PyEBSDIndex on additively manufactured AlSi10Mg) that needs the downstream grain-characterization stage to run without MATLAB.", + "started_at": "2026-07-29T03:32:11.680789Z", + "crow": "job-futurehouse-paperqa3", + "notification_enabled": false, + "notification_type": null, + "continued_trajectory_id": null, + "is_trajectory_owner": true, + "gcloud_operation_name": "sandbox://f55d0bdd-87d8-4cf2-8d86-23309987262a/sandbox-claim-5fbcc685/job-futurehouse-paperqa3-agent-warmpool-bdnd8", + "task_summary": "Comparison of Python Native Open Source EBSD Tools For Grain Reconstruction Texture And Misorientation Metrics", + "id": "f55d0bdd-87d8-4cf2-8d86-23309987262a" +} \ No newline at end of file diff --git a/scripts/dictionary_indexing_test.py b/scripts/dictionary_indexing_test.py new file mode 100644 index 0000000..ee29a65 --- /dev/null +++ b/scripts/dictionary_indexing_test.py @@ -0,0 +1,233 @@ +# Dictionary indexing test on AlSi10Mg Area 2, first 25 map rows (4,900 patterns). +# +# Compares Hough indexing (PyEBSDIndex) against dictionary indexing + orientation +# refinement (kikuchipy, EMsoft-simulated master patterns from Zenodo), and runs +# an Al-vs-Si dictionary phase-discrimination test that Hough cannot do. +# +# Free stack: pip install kikuchipy pyebsdindex orix matplotlib +# +# Data: EDAX .up2 v3, 42-byte header, 111x111 px patterns, 196x446 map, 2 um step. +# Only the first 25 rows are fetched (HTTP range request against the public Box share). + +import gzip +import json +import shutil +import time +import urllib.request +from pathlib import Path + +import matplotlib + +matplotlib.use("Agg") +import matplotlib.pyplot as plt +import numpy as np + +import kikuchipy as kp +from orix import io, plot, sampling +from orix.crystal_map import CrystalMap +from orix.quaternion import Orientation, Rotation, symmetry + +N_ROWS, N_COLS, PAT = 25, 196, 111 +OUT = Path(__file__).resolve().parents[1] / "example" / "area2-results" / "dictionary-indexing" +OUT.mkdir(parents=True, exist_ok=True) + +# ---------------------------------------------------------------- 1. get the patterns +UP2_URL = ( + "https://byu.app.box.com/index.php?rm=box_download_shared_file" + "&shared_name=hz48nsx7ciln1jifm3u0yo5vqpge2lma&file_id=f_2371754519702" +) +slice_file = Path("/tmp/area2_slice.bin") +if not slice_file.exists(): + n_bytes = 42 + N_ROWS * N_COLS * PAT * PAT * 2 + req = urllib.request.Request(UP2_URL, headers={"Range": "bytes=0-%d" % (n_bytes - 1)}) + with urllib.request.urlopen(req) as r, open(slice_file, "wb") as f: + shutil.copyfileobj(r, f) + +raw = np.fromfile(slice_file, dtype=np.uint16, offset=42).reshape(N_ROWS, N_COLS, PAT, PAT) + +# ---------------------------------------------------------------- 2. background correction +s = kp.signals.EBSD(raw.copy()) +s.static_background = raw.mean(axis=(0, 1)).astype(np.uint16) # .up2 stores no flat-field +s.remove_static_background(operation="subtract", show_progressbar=False) +s.remove_dynamic_background(show_progressbar=False) +corrected = s.data.copy() + +# ---------------------------------------------------------------- 3. Hough baseline (TEAM-style) +al_mp = kp.data.ebsd_master_pattern( + "al", allow_download=True, show_progressbar=False, + projection="lambert", energy=20, hemisphere="upper", +) +si_mp = kp.data.ebsd_master_pattern( + "si", allow_download=True, show_progressbar=False, + projection="lambert", energy=20, hemisphere="upper", +) + +from orix.crystal_map import PhaseList + +al_phases = PhaseList(al_mp.phase) +det0 = kp.detectors.EBSDDetector((PAT, PAT), sample_tilt=70.0) +indexer = det0.get_indexer(al_phases, nBands=9, tSigma=2, rSigma=2) +s_grid, _ = s.extract_grid((5, 5), return_indices=True) +det = s_grid.hough_indexing_optimize_pc( + pc0=[0.5, 0.5, 0.6], indexer=indexer, batch=False, method="PSO", search_limit=0.2 +) +indexer = det.get_indexer(al_phases, nBands=9, tSigma=2, rSigma=2) +t0 = time.time() +xmap_hough = s.hough_indexing(phase_list=al_phases, indexer=indexer, verbose=0) +t_hough = time.time() - t0 +g_hough = xmap_hough.rotations # flat, map (row-major) order + +# ---------------------------------------------------------------- 4. boost SNR for DI +# Patterns are too noisy for per-pixel cross-correlation; average each pattern +# with its neighbours (grains are tens of pixels, so no orientation mixing). +s.average_neighbour_patterns(show_progressbar=False) + +signal_mask = ~kp.filters.Window("circular", (PAT, PAT)).astype(bool) # True = excluded + +# ---------------------------------------------------------------- 5. dictionary indexing, Al and Si +R = sampling.get_sample_fundamental(method="cubochoric", resolution=2.5, point_group=symmetry.Oh) +yy, xx = np.indices((N_ROWS, N_COLS)) +results = {} +for name, mp in [("al", al_mp), ("si", si_mp)]: + t0 = time.time() + sim = mp.get_patterns( + rotations=R, detector=det, energy=20, compute=True, + show_progressbar=False, dtype_out=np.float32, + ) + t_proj = time.time() - t0 + t0 = time.time() + xmap_di = s.dictionary_indexing(sim, metric="ncc", keep_n=5, signal_mask=signal_mask) + t_di = time.time() - t0 + xmap_best = CrystalMap( + rotations=Rotation(xmap_di.rotations[:, 0].data), + x=xx.ravel().astype(float), y=yy.ravel().astype(float), + phase_list=xmap_di.phases, + ) + t0 = time.time() + xmap_ref = s.refine_orientation( + xmap=xmap_best, detector=det, master_pattern=mp, energy=20, + signal_mask=signal_mask, method="minimize", trust_region=[2, 2, 2], + ) + t_ref = time.time() - t0 + results[name] = { + "xmap_di": xmap_di, "xmap_ref": xmap_ref, + "t_proj": t_proj, "t_di": t_di, "t_ref": t_ref, + } + del sim + +ncc_al_di = results["al"]["xmap_di"].scores[:, 0] +ncc_al = results["al"]["xmap_ref"].scores.flatten() +ncc_si = results["si"]["xmap_ref"].scores.flatten() +g_di = results["al"]["xmap_ref"].rotations + +# ---------------------------------------------------------------- 6. compare with Hough +O_h = Orientation(g_hough, symmetry.Oh) +O_d = Orientation(g_di, symmetry.Oh) +misori = O_h.angle_with(O_d, degrees=True) +ci = xmap_hough.cm +fit = xmap_hough.fit +low_ci = ci < 0.1 + +stats = { + "n_patterns": int(misori.size), + "dictionary_size": int(R.size), + "dictionary_resolution_deg": 2.5, + "pc_bruker": [round(float(v), 4) for v in det.pc_average], + "hough": { + "s_per_map": round(t_hough, 1), + "ci_gt_0.1": round(float((ci > 0.1).mean()), 4), + "median_ci": round(float(np.median(ci)), 3), + "median_fit_deg": round(float(np.median(fit)), 3), + }, + "di_al": { + "s_projection": round(results["al"]["t_proj"], 1), + "s_indexing": round(results["al"]["t_di"], 1), + "s_refinement": round(results["al"]["t_ref"], 1), + "median_ncc_di": round(float(np.median(ncc_al_di)), 3), + "median_ncc_refined": round(float(np.median(ncc_al)), 3), + }, + "agreement_hough_vs_di": { + "median_misorientation_deg": round(float(np.median(misori)), 2), + "frac_within_2deg": round(float((misori < 2).mean()), 3), + "frac_within_5deg": round(float((misori < 5).mean()), 3), + "median_misori_where_ci_lt_0.1": round(float(np.median(misori[low_ci])), 2) if low_ci.any() else None, + "n_ci_lt_0.1": int(low_ci.sum()), + }, + "phase_test_al_vs_si": { + "median_ncc_al": round(float(np.median(ncc_al)), 3), + "median_ncc_si": round(float(np.median(ncc_si)), 3), + "frac_al_wins": round(float((ncc_al > ncc_si).mean()), 4), + "median_delta_ncc": round(float(np.median(ncc_al - ncc_si)), 3), + }, +} +(OUT / "stats.json").write_text(json.dumps(stats, indent=2)) +print(json.dumps(stats, indent=2)) + +# ---------------------------------------------------------------- 7. plots +mis2d = misori.reshape(N_ROWS, N_COLS) +ncc2d = ncc_al.reshape(N_ROWS, N_COLS) + +fig, ax = plt.subplots(2, 2, figsize=(14, 5.2)) +im = ax[0, 0].imshow(ncc_al_di.reshape(N_ROWS, N_COLS), cmap="viridis") +ax[0, 0].set_title("DI best NCC (before refinement)") +fig.colorbar(im, ax=ax[0, 0], shrink=0.8) +im = ax[0, 1].imshow(ncc2d, cmap="viridis") +ax[0, 1].set_title("NCC after refinement") +fig.colorbar(im, ax=ax[0, 1], shrink=0.8) +im = ax[1, 0].imshow(mis2d, cmap="inferno", vmax=10) +ax[1, 0].set_title("misorientation DI vs Hough (deg, capped 10)") +fig.colorbar(im, ax=ax[1, 0], shrink=0.8) +im = ax[1, 1].imshow(ci.reshape(N_ROWS, N_COLS), cmap="viridis") +ax[1, 1].set_title("Hough CI") +fig.colorbar(im, ax=ax[1, 1], shrink=0.8) +for a in ax.ravel(): + a.set_xticks([]), a.set_yticks([]) +plt.tight_layout() +plt.savefig(OUT / "di_quality_maps.png", dpi=130) + +ckey = plot.IPFColorKeyTSL(symmetry.Oh) +fig, ax = plt.subplots(2, 1, figsize=(14, 4.6)) +ax[0].imshow(ckey.orientation2color(O_h).reshape(N_ROWS, N_COLS, 3)) +ax[0].set_title("IPF-Z, Hough") +ax[1].imshow(ckey.orientation2color(O_d).reshape(N_ROWS, N_COLS, 3)) +ax[1].set_title("IPF-Z, dictionary indexing + refinement") +for a in ax: + a.set_xticks([]), a.set_yticks([]) +plt.tight_layout() +plt.savefig(OUT / "ipf_hough_vs_di.png", dpi=130) + +fig, ax = plt.subplots(1, 2, figsize=(12, 4)) +ax[0].hist(ncc_al_di, bins=80, alpha=0.6, label="DI (2.5 deg dictionary)") +ax[0].hist(ncc_al, bins=80, alpha=0.6, label="after refinement") +ax[0].set_xlabel("NCC"), ax[0].legend(), ax[0].set_title("Al: match quality") +ax[1].hist(ncc_al - ncc_si, bins=80, color="tab:green") +ax[1].axvline(0, color="k", lw=0.8) +ax[1].set_xlabel("NCC(Al) - NCC(Si)") +ax[1].set_title("phase discrimination: Al vs Si dictionaries") +plt.tight_layout() +plt.savefig(OUT / "ncc_histograms.png", dpi=130) + +# example pattern: raw vs averaged vs best-match simulation +r, c = 10, 100 +i = r * N_COLS + c +best_sim = al_mp.get_patterns( + rotations=Rotation(g_di[i].data), detector=det, energy=20, + compute=True, show_progressbar=False, dtype_out=np.float32, +).data[0] +fig, ax = plt.subplots(1, 4, figsize=(16, 4.2)) +for a, (im_, t) in zip(ax, [ + (raw[r, c], "raw pattern (%d,%d)" % (r, c)), + (corrected[r, c], "background-corrected"), + (s.data[r, c], "+ neighbour averaging"), + (best_sim, "best-match simulation\nNCC=%.2f" % ncc_al[i]), +]): + a.imshow(im_, cmap="gray"), a.set_title(t), a.axis("off") +plt.tight_layout() +plt.savefig(OUT / "pattern_vs_simulation.png", dpi=130) + +# ---------------------------------------------------------------- 8. export refined map +ang_path = OUT / "area2_slice_di_refined.ang" +io.save(str(ang_path), results["al"]["xmap_ref"], overwrite=True) +with open(ang_path, "rb") as f_in, gzip.open(str(ang_path) + ".gz", "wb") as f_out: + shutil.copyfileobj(f_in, f_out) +ang_path.unlink() diff --git a/scripts/dream3dnx_grain_reconstruction.py b/scripts/dream3dnx_grain_reconstruction.py new file mode 100644 index 0000000..26312d6 --- /dev/null +++ b/scripts/dream3dnx_grain_reconstruction.py @@ -0,0 +1,208 @@ +"""DREAM3D-NX (simplnx) grain reconstruction parameter sweeps on the Area 2 map. + +Input: example/area2-results/area2_full.ang (gunzip area2_full.ang.gz first). +The kikuchipy .ang writer zeroed the CI/IQ columns, so the pre-segmentation +quality filter gates on the pattern fit column (degrees; lower is better). + +Run inside the conda env: + conda create -y -n nx -c bluequartzsoftware -c conda-forge python=3.12 dream3dnx numpy matplotlib-base + conda run -n nx python scripts/dream3dnx_grain_reconstruction.py +""" + +import json +from pathlib import Path + +import matplotlib.pyplot as plt +import numpy as np +import simplnx as nx +import orientationanalysis as oa + +ANG = Path("example/area2-results/area2_full.ang") +OUT = Path("example/area2-results/grain-reconstruction") +OUT.mkdir(parents=True, exist_ok=True) + +GEOM = nx.DataPath("Geom") +CELL = "Geom/Cell Data/" +STEP_UM = 2.0 # map step size -> each pixel is 2x2 um + + +def reconstruct(mis_tol=5.0, min_size=10, fit_max=1.4, erode_iters=0, fill_defect=0): + """One DREAM3D-NX reconstruction; returns the FeatureIds image and stats.""" + ds = nx.DataStructure() + + # 1. Read orientations from the .ang produced by the kikuchipy Hough run + oa.ReadAngDataFilter.execute( + data_structure=ds, input_file=str(ANG), output_image_geometry_path=GEOM) + + # 2. Pre-segmentation quality filter: keep pixels with fit < fit_max (deg) + if fit_max is not None: + t = nx.ArrayThreshold() + t.array_path = nx.DataPath(CELL + "Fit") + t.comparison = nx.ArrayThreshold.ComparisonType.LessThan + t.value = fit_max + ts = nx.ArrayThresholdSet() + ts.thresholds = [t] + nx.MultiThresholdObjectsFilter.execute( + data_structure=ds, array_thresholds_object=ts, + created_mask_type=nx.DataType.boolean, output_data_array_name="Mask") + + # 3. Euler angles -> quaternions (what the segmenter compares) + oa.ConvertOrientationsFilter.execute( + data_structure=ds, + input_orientation_array_path=nx.DataPath(CELL + "EulerAngles"), + input_representation_index=0, # Euler + output_orientation_array_name="Quats", + output_representation_index=2) # Quaternion + + # 4. Grain segmentation: flood-fill, boundary = misorientation > mis_tol + res = oa.EBSDSegmentFeaturesFilter.execute( + data_structure=ds, + input_image_geometry_path=GEOM, + cell_quats_array_path=nx.DataPath(CELL + "Quats"), + cell_phases_array_path=nx.DataPath(CELL + "Phases"), + crystal_structures_array_path=nx.DataPath("Geom/Cell Ensemble Data/CrystalStructures"), + cell_mask_array_path=nx.DataPath(CELL + "Mask"), + use_mask=fit_max is not None, + misorientation_tolerance=mis_tol, + randomize_features=True) + + assert not res.errors, res.errors + fids_path = nx.DataPath(CELL + "FeatureIds") + n_raw_grains = int(ds[fids_path].store.npview().max()) + + # 5. Minimum-points-per-grain cutoff (needs NumElements per feature first) + if min_size > 1: + nx.ComputeFeatureSizesFilter.execute( + data_structure=ds, input_image_geometry_path=GEOM, + feature_ids_path=fids_path, + feature_attribute_matrix_path=nx.DataPath("Geom/Cell Feature Data")) + res = nx.RequireMinimumSizeFeaturesFilter.execute( + data_structure=ds, input_image_geometry_path=GEOM, + feature_ids_path=fids_path, + num_cells_path=nx.DataPath("Geom/Cell Feature Data/NumElements"), + min_allowed_features_size=min_size) + assert not res.errors, res.errors # errors at e.g. cutoff > largest grain + + unassigned_before_cleanup = float( + (ds[fids_path].store.npview() == 0).mean()) + + # 6. Cleanup: erode bad data (operation_index 1) grows grains into holes; + # FillBadData replaces enclosed hole regions smaller than fill_defect px + if erode_iters > 0: + nx.ErodeDilateBadDataFilter.execute( + data_structure=ds, input_image_geometry_path=GEOM, + feature_ids_path=fids_path, + num_iterations=erode_iters, operation_index=1) + if fill_defect > 0: + nx.FillBadDataFilter.execute( + data_structure=ds, input_image_geometry_path=GEOM, + feature_ids_path=fids_path, + cell_phases_array_path=nx.DataPath(CELL + "Phases"), + min_allowed_defect_size=fill_defect) + + fids = ds[fids_path].store.npview()[0, :, :, 0].copy() + + sizes_px = np.bincount(fids.ravel())[1:] # id 0 = unassigned + sizes_px = sizes_px[sizes_px > 0] + eq_diam_um = 2.0 * np.sqrt(sizes_px * STEP_UM**2 / np.pi) + stats = { + "n_raw_grains": n_raw_grains, + "n_grains": int(sizes_px.size), + "retained_frac": float((fids > 0).mean()), + "cleanup_altered_frac": float(unassigned_before_cleanup - (fids == 0).mean()), + "mean_diam_um": float(eq_diam_um.mean()), + "area_weighted_mean_diam_um": float((sizes_px * eq_diam_um).sum() / sizes_px.sum()), + "speckle_grain_frac": float((sizes_px < 10).mean()), + "largest_grain_px": int(sizes_px.max()), + } + return fids, stats + + +def render(panels, fname, suptitle): + """panels: list of (featureIds image, title).""" + rng = np.random.default_rng(7) + fig, axes = plt.subplots(1, len(panels), figsize=(3.1 * len(panels), 7.5)) + for ax, (fids, title) in zip(np.atleast_1d(axes), panels): + colors = rng.uniform(0.15, 1.0, (fids.max() + 1, 3)) + colors[0] = 0.0 # unassigned pixels = black + ax.imshow(colors[fids], interpolation="nearest") + ax.set_title(title, fontsize=9) + ax.set_xticks([]), ax.set_yticks([]) + fig.suptitle(suptitle, fontsize=11, y=0.995) + fig.tight_layout(rect=(0, 0, 1, 0.93)) + fig.savefig(OUT / fname, dpi=150) + plt.close(fig) + + +CONFIGS = { + # baseline + "baseline": dict(mis_tol=5.0, min_size=10, fit_max=1.4), + # A: misorientation threshold defining a boundary + "tol_1deg": dict(mis_tol=1.0, min_size=10, fit_max=1.4), + "tol_2deg": dict(mis_tol=2.0, min_size=10, fit_max=1.4), + "tol_15deg": dict(mis_tol=15.0, min_size=10, fit_max=1.4), + # B: minimum-points-per-grain cutoff + "min_1px": dict(mis_tol=5.0, min_size=1, fit_max=1.4), + "min_100px": dict(mis_tol=5.0, min_size=100, fit_max=1.4), + # note: cutoffs above the largest raw feature (493 px) error out and no-op + "min_400px": dict(mis_tol=5.0, min_size=400, fit_max=1.4), + # C: pre-segmentation quality filter (pattern fit, degrees) + "filter_none": dict(mis_tol=5.0, min_size=10, fit_max=None), + "filter_strict": dict(mis_tol=5.0, min_size=10, fit_max=0.5), + # D: cleanup passes on the strict-filter map (half the pixels are holes) + "strict_erode2": dict(mis_tol=5.0, min_size=10, fit_max=0.5, erode_iters=2), + "strict_erode10": dict(mis_tol=5.0, min_size=10, fit_max=0.5, erode_iters=10), + "strict_fill": dict(mis_tol=5.0, min_size=10, fit_max=0.5, fill_defect=1000), + # worse-then-better demonstration + "worst": dict(mis_tol=1.0, min_size=1, fit_max=None), + "recovered": dict(mis_tol=5.0, min_size=10, fit_max=1.4, erode_iters=2), +} + +results, images = {}, {} +for name, cfg in CONFIGS.items(): + fids, stats = reconstruct(**cfg) + results[name] = {"config": {k: (v if v is not None else "off") for k, v in cfg.items()}, **stats} + images[name] = fids + print(f"{name:16s} grains={stats['n_grains']:6d} retained={stats['retained_frac']:.3f} " + f"mean_d={stats['mean_diam_um']:.1f}um speckle={stats['speckle_grain_frac']:.2f}") + +with open(OUT / "stats.json", "w") as f: + json.dump(results, f, indent=2) + + +def title(name, label): + s = results[name] + return (f"{label}\n{s['n_raw_grains']:,} raw / {s['n_grains']:,} kept grains\n" + f"retained {s['retained_frac']:.0%} | mean {s['mean_diam_um']:.1f} um | " + f"AW {s['area_weighted_mean_diam_um']:.0f} um") + + +render([(images[n], title(n, l)) for n, l in + [("tol_1deg", "1 deg"), ("tol_2deg", "2 deg"), ("baseline", "5 deg (baseline)"), ("tol_15deg", "15 deg")]], + "sweep_misorientation_tolerance.png", + "Misorientation tolerance (min 10 px, fit < 1.4 deg, no cleanup)") + +render([(images[n], title(n, l)) for n, l in + [("min_1px", "1 px (off)"), ("baseline", "10 px (baseline)"), ("min_100px", "100 px"), ("min_400px", "400 px")]], + "sweep_min_grain_size.png", + "Minimum-points-per-grain cutoff (5 deg, fit < 1.4 deg, no cleanup)") + +render([(images[n], title(n, l)) for n, l in + [("filter_none", "no filter"), ("baseline", "fit < 1.4 deg (baseline)"), ("filter_strict", "fit < 0.5 deg")]], + "sweep_quality_filter.png", + "Pre-segmentation quality filter (5 deg, min 10 px, no cleanup)") + +render([(images[n], title(n, l)) for n, l in + [("filter_strict", "no cleanup"), ("strict_erode2", "erode bad data x2"), + ("strict_erode10", "erode bad data x10"), ("strict_fill", "fill defects < 1000 px")]], + "sweep_cleanup.png", + "Cleanup passes on the fit < 0.5 deg map (49% of pixels start as holes)") + +render([(images[n], title(n, l)) for n, l in + [("worst", "WORST: 1 deg, no min size,\nno filter, no cleanup"), + ("baseline", "baseline: 5 deg, min 10 px,\nfit < 1.4 deg"), + ("recovered", "recovered: baseline\n+ erode bad data x2")]], + "worse_then_better.png", + "Deliberately degraded vs recovered reconstruction") + +print("done ->", OUT) diff --git a/scripts/master_pattern_python_test.py b/scripts/master_pattern_python_test.py new file mode 100644 index 0000000..c9f5fa7 --- /dev/null +++ b/scripts/master_pattern_python_test.py @@ -0,0 +1,81 @@ +"""Pure-Python (kinematical) vs EMsoft (dynamical) master patterns for Al. + +Computes a kinematical EBSD master pattern for fcc Al entirely in Python +(kikuchipy + diffsims, no Fortran), downloads the EMsoft-simulated dynamical +master pattern from Zenodo, and compares both on the sphere and projected +onto the Area 2 detector geometry for the same orientation. + +Run: python scripts/master_pattern_python_test.py +Requires: pip install kikuchipy diffsims matplotlib +""" + +import time + +import matplotlib.pyplot as plt +import numpy as np +from diffpy.structure import Atom, Lattice, Structure +from diffsims.crystallography import ReciprocalLatticeVector +from orix.crystal_map import Phase +from orix.quaternion import Rotation +import kikuchipy as kp + +OUT_DIR = "example/area2-results/master-pattern-comparison" + +# --- Kinematical master pattern, pure Python ------------------------------- +phase = Phase( + name="al", + space_group=225, + structure=Structure( + atoms=[Atom("Al", [0, 0, 0])], + lattice=Lattice(4.05, 4.05, 4.05, 90, 90, 90), + ), +) +t0 = time.time() +ref = ReciprocalLatticeVector.from_min_dspacing(phase, 0.7) +ref.sanitise_phase() +ref.calculate_structure_factor() +ref.calculate_theta(20e3) +simulator = kp.simulations.KikuchiPatternSimulator(ref) +mp_kin = simulator.calculate_master_pattern(half_size=200, hemisphere="upper") +t_kin = time.time() - t0 +print(f"Kinematical master pattern: {ref.size} reflectors, {t_kin:.1f} s") + +# --- Dynamical master pattern, EMsoft via Zenodo --------------------------- +mp_dyn = kp.data.ebsd_master_pattern( + "al", allow_download=True, projection="stereographic", energy=20, + hemisphere="upper", +) + +# --- Project both onto the Area 2 detector for one orientation ------------- +det = kp.detectors.EBSDDetector( + (111, 111), sample_tilt=70.0, pc=(0.518, 0.336, 0.753), convention="edax", +) +rot = Rotation.from_euler(np.deg2rad([30, 45, 60])) + +mp_dyn_lam = kp.data.ebsd_master_pattern( + "al", allow_download=True, projection="lambert", energy=20, + hemisphere="upper", +) +mp_dyn_lam.phase = phase +pat_dyn = mp_dyn_lam.get_patterns(rot, det, energy=20, compute=True) + +mp_kin_lam = mp_kin.as_lambert() +mp_kin_lam.phase = phase +pat_kin = mp_kin_lam.get_patterns(rot, det, energy=20, compute=True) + +# --- Figure ---------------------------------------------------------------- +fig, axes = plt.subplots(1, 4, figsize=(16, 4.4)) +panels = [ + (mp_dyn.data, "Dynamical master pattern\n(EMsoft, Zenodo download)"), + (mp_kin.data, f"Kinematical master pattern\n(pure Python, {t_kin:.0f} s)"), + (pat_dyn.data.squeeze(), "Detector pattern, dynamical\n(Area 2 geometry)"), + (pat_kin.data.squeeze(), "Detector pattern, kinematical\n(same orientation)"), +] +for ax, (img, title) in zip(axes, panels): + ax.imshow(img, cmap="gray") + ax.set_title(title, fontsize=10) + ax.axis("off") +fig.tight_layout() +fig.savefig(f"{OUT_DIR}/kinematical_vs_dynamical_al.png", dpi=150, + bbox_inches="tight") +print(f"Saved {OUT_DIR}/kinematical_vs_dynamical_al.png")