A comprehensive case study and research paper implementation utilizing deep learning techniques to accurately classify ECG images into five distinct medical categories. Developed with VGG-19 and advanced CNN architectures, this solution is seamlessly deployed via Streamlit, featuring a professional and user-friendly interface for enhanced accessibility and efficiency.
This project explores the power of deep learning in ECG image classification, providing an AI-driven approach for cardiac and COVID-related diagnostics.
- Implement and evaluate CNN-based models on ECG image data.
- Utilize image preprocessing techniques (OpenCV) to enhance signal clarity.
- Develop an interactive web application using Streamlit for healthcare professionals.
- Demonstrate the impact of deep learning in medical image diagnostics.
- Advanced Deep Learning Models: Built using ANN, CNN, and VGG-19 for ECG classification.
- Data Preprocessing: Improved model performance with SMOTE and OpenCV-based enhancement techniques.
- End-to-End Pipeline: Data → Model → Deployment, streamlining AI-powered workflows.
- Interactive Web App: Real-time ECG classification deployed via Streamlit with a professional UI.
- Performance Analysis: Comparative study between ANN and CNN models, proving CNN’s superiority for image-based classification.
- Dataset Source: ECG Images dataset of Cardiac and COVID-19 Patients
- DOI: 10.17632/gwbz3fsgp8.1
| Model | Accuracy | Notes |
|---|---|---|
| ANN | 20.13% | Poor results, proves ANN is not suited for raw image data |
| CNN | 90.45% | Excellent performance, slight overfitting |
| VGG-19 | 86.38% | Strong balance between accuracy and generalization |
✅ This practical comparison proves CNNs are far superior to ANN for image-based classification.
Rupak C. Jogi
Data Scientist & Analyst | AI/ML | Deep Learning Enthusiast
# 1. Clone the repository
git clone https://github.com/RCJ360/ECG-Image-Classification-App.git
cd ecg-image-analyzer
# 2. Install dependencies
pip install -r requirements.txt
# 3. Run the app
streamlit run app.py