CoronaryAI is a medical-grade computer vision platform designed to detect blockages in coronary angiograms. It combines an optimized RF+SVM Ensemble machine learning pipeline with clinical-grade Explainable AI (XAI) visualizations and automated medical reporting.
1. Intelligent Landing Page
2. High-Precision Scan & Analysis
Real-time vessel density heatmap and electric blue clinical markers pinpointing potential occlusions.
3. Clinical Reporting & Diagnostics
4. Performance Dashboard
Uses Frangi Vesselness Filters to reconstruct arterial paths. The system overlays a high-contrast INFERNO heatmap that visualizes vessel density and highlights structural deviations in real-time.
Utilizes a Gaussian Centrality Ranking algorithm to identify the primary site of occlusion. It places an Electric Blue Clinical Marker on the most critical segment, ensuring radiologists stay focused on relevant pathology.
Generate formal, hospital-ready diagnostic reports with a single click. Includes:
- High-fidelity scans with marked highlights.
- Comparative model confidence gauges.
- AI-generated Clinical Interpretation and Action Plans.
A built-in Diagnostic Assistant (powered by Groq/LLaMA 3) fuzes the model's raw 15-dimensional probability space into human-readable medical insights.
The system uses a weighted voting ensemble for maximum reliability on medical datasets:
- Random Forest (RF): Selected for its robustness against noisy texture data in angiograms.
- Support Vector Machine (SVM): Uses a Radial Basis Function (RBF) kernel to separate high-dimensional GLCM and LBP feature vectors.
- Benchmark Performance: 70.4% Accuracy on cross-validation sets.
- Reconstructed Vessel Area: Extracted via Bilateral Blur + CLAHE + Frangi.
- GLCM Textures (4): Contrast, Energy, Homogeneity, and Correlation (captures plaque density).
- LBP Histograms (10): 10-bin Local Binary Patterns for pixel-level micro-texture analysis.
We have provided professional startup scripts that launch both the Backend and Frontend with a single command.
For Mac/Linux:
./start.shFor Windows:
Start_CoronaryAI.batIf you prefer to run the services individually:
1. Backend:
cd backend
source venv/bin/activate # (or venv\Scripts\activate on Windows)
PORT=5001 python3 app.py2. Frontend:
cd frontend
npm run devNavigate to http://localhost:5173 in your browser.
├── backend/
│ ├── model/ # ML Core (RF, SVM, Feature extraction)
│ │ ├── saved/ # Persisted .pkl models & scalers
│ │ ├── predict.py # Synthesis & XAI Logic
│ │ └── train.py # Training Ensemble pipeline
│ └── app.py # Flask REST API
├── frontend/
│ ├── src/
│ │ ├── components/ # Dashboard & Chat components
│ │ ├── pages/ # Predict, Home, Statistics
│ │ └── index.css # Premium Glassmorphic Design System
│ └── package.json # Node Dependencies (jsPDF, html2canvas)
This tool is for research and educational purposes only. It is not a substitute for professional medical diagnosis. Always consult a qualified cardiologist for clinical decisions.
Developed with ❤️ by mehakmeet for Modern Cardiology