LVMC-HeartSeg

Automatic segmentation of cardiac substructures (left ventricular myocardial walls and coronary arteries)
on non-angiographic contrast CT (cCT), based on nnU-Net v2.

---

🧩 Features

• nnU-Net v2–based 3D full-resolution architecture for non-angiographic cCT
• Segmentation of LV myocardial walls (apical, septal, inferior, anterior, lateral), coronary arteries (LMCA, LAD, LCx, RCA), and major chambers/vessels
• Iterative AI-assisted annotation workflow with human-in-the-loop refinement
• Complete hyperparameter configurations in configs/training_hyperparameters.yaml

---

⚙️ Environment Setup

conda env create -f environment.yml
conda activate radai

---

🚀 Usage

python train.py --config configs/training_hyperparameters.yaml
python inference.py --input data/example_case/CT.nii.gz --output results/LVMC_mask.nii.gz

---

🧠 Citation

If you use LVMC-HeartSeg in your research or publications, please cite:

Wei J, Knoll M, Furkel J, Abdollahi A, et al.
LVMC-HeartSeg: Automatic left ventricular myocardial and coronary artery segmentation on non-angiographic cCT using nnU-Net v2.
2025. Manuscript in preparation.
https://github.com/jielinwei/LVMC-HeartSeg

@software{Wei2025_LVMC_HeartSeg,
  author  = {Wei, Jielin and Knoll, Maximilian and Furkel, Jennifer and Abdollahi, Amir},
  title   = {LVMC-HeartSeg: Automatic LV myocardial wall and coronary artery segmentation on non-angiographic cCT using nnU-Net v2},
  year    = {2025},
  url     = {https://github.com/jielinwei/LVMC-HeartSeg},
  note    = {Code released under CC BY-NC-SA 4.0 for academic use}
}

---

🧩 License

This project is licensed under CC BY-NC-SA 4.0 (academic and research use only).
Commercial use is not permitted without explicit permission.
License: https://creativecommons.org/licenses/by-nc-sa/4.0/
Commercial contact: jielin.wei@dkfz-heidelberg.de

---

🔒 Intended Use & Data Privacy

Intended Use.
This repository is provided for academic research and method reproduction only.
It is not intended for clinical diagnosis, treatment planning, or any other clinical decision-making.

Data Privacy.
This repository does not include any protected health information (PHI).
The directory data/example_case/ is reserved for anonymized demo samples only.
Do not upload or share real patient data through this repository.

---

🧑‍💻 Contact

Jielin Wei, MD
Postdoctoral Researcher, DKFZ/NCT Heidelberg
📧 jielin.wei@dkfz-heidelberg.de