Automated Stenosis Detection: Classical Image Processing Pipeline

A self-contained Jupyter notebook for detecting artery stenosis from angiography images using classical image-processing techniques: denoising, edge enhancement, rotation alignment, and diameter estimation.

Repository Structure

.
├── README.md
├── notebooks/
│   └── stenosis_detection.ipynb       ← Jupyter notebook with full code & explanations
├── data/
│   └── angio_n.bmp                    ← Sample angiography image
├── results/
│   ├── stenosis_crop.png             ← Binarized cropped vessel region
│   └── stenosis_summary.png          ← Final summary (diameter & stenosis %)
└── requirements.txt

📂 Requirements

• Python 3.7+
• NumPy
• OpenCV
• scikit-image
• Matplotlib
• JupyterLab or Notebook

Install dependencies:

pip install -r requirements.txt

🚀 Usage

1. Clone the repository

   git clone https://github.com/PixelErfan/automated-stenosis-detection.git
   cd automated-stenosis-detection

2. Launch the Jupyter notebook

   jupyter notebook notebooks/stenosis_detection.ipynb

3. Run all cells, step-by-step:

   1. Load and display angiography image
   2. Denoise (median, morphological)
   3. Enhance edges with gradient & Laplacian
   4. Detect edges using Canny
   5. Automatically rotate vessel
   6. Binarize and crop region of interest
   7. Estimate maximum/minimum diameter
   8. Compute stenosis percentage
   9. Save result figures

🧠 Method Overview

The notebook uses a fully classical (non-deep learning) approach to analyze blood vessel structure:

1. Denoising with filters to reduce noise artifacts
2. Edge Enhancement to highlight vessel boundaries
3. Auto-Rotation to align vessel horizontally
4. Width Estimation by scanning white pixel runs in each row
5. Stenosis Calculation using:
   Stenosis (%) = (1 - D_stenosis / D_normal) × 100

📈 Results

• Normal diameter (max): 58 px
• Stenotic diameter (min): 1 px
• Percent stenosis: 98.28 %

🖼️ Visual Output

Input Angiography Image

Cropped Binary Vessel Mask

Summary Image

📄 License

This project is released under the MIT License. See LICENSE for details.

👤 Author

Mohammad Erfan Hasani · m_erfan_h78@yahoo.com
GitHub: @PixelErfan