Coronary Artery Stenosis Detection and Classification

Overview

This AI-powered project focuses on the automatic detection and classification of coronary artery stenosis from angiographic X-ray images. By combining deep learning models for both Detection and classification of Stenosis, the system supports clinicians in accurate, early diagnosis of CAD (Coronary Artery Disease).

---

Model Architexture

---

Screenshots of the Interface

---

🧠 Features

• Automated Stenosis Detection using YOLOv8
• Artery Classification into LAD, LCX, and RCA using an ensemble of ResNet50, VGG16, and EfficientNetB0
• Preprocessing: CLAHE, denoising, and augmentation to improve sensitivity
• XAI Support: Explainable AI tools for better clinical transparency

---

📊 Results

• Ensemble Accuracy: 99.29% (Classification of Stenosis)
• Improved vessel-specific precision and recall

---

📂 Dataset

• Name: ARCADE (Automatic Region-based Coronary Artery Disease diagnostics)
• Format: .png images with .json annotations for stenosis and vessel labels

---

🛠️ Tech Stack

• Python 3.10+
• PyTorch, YOLOv8, EfficientNet, ResNet, VGG
• OpenCV, NumPy, Pandas, Scikit-learn
• Jupyter Notebook for development and experiments

---

🌍 Impact

• Supports SDG 3: Good Health and Well-being
• Reduces human error in diagnostics
• Enables real-time, explainable, and accurate detection of CAD

---

🚀 How to Run

Follow these steps to set up and run the coronary artery stenosis detection and classification system:

1. Clone the Repository

git clone https://github.com/paranthagan78/Stenosis-Detection-and-Classification.git
cd Stenosis-Detection-and-Classification

---

2. Set Up the Environment

It’s recommended to use a virtual environment (optional but clean):

For venv:

python -m venv venv
venv\Scripts\activate  # On Windows
# OR
source venv/bin/activate  # On macOS/Linux

---

3. Install Dependencies

pip install -r requirements.txt

---

4. Prepare the Dataset

• Download the ARCADE dataset from: 🔗 https://zenodo.org/records/10390295

• Place the dataset folders (stenosis/ and syntax/) in the appropriate directories expected by your code.

• Ensure the structure includes:

  • .png images
  • .json annotation files (with bounding boxes and vessel labels)

---

5. Run Detection and Classification Notebooks

Open Jupyter Notebook or VS Code and run the following notebooks in order:

• Detection_Stenosis/yolov8_train_detect.ipynb – Train or infer stenosis detection

• Classification_Stenosis/final_class_ensemble.ipynb – Classify affected vessels using ensemble model

• Optionally run:

  • Classification_Stenosis/auc_roc.ipynb
  • Classification_Stenosis/final_class_resnet.ipynb
  • Classification_Stenosis/final_class_vgg.ipynb

---

6. View Results

• Check output directories or notebook visualizations for:

  • Detected stenotic regions
  • Predicted artery classes (LAD, LCX, RCA)
  • Confusion matrices and performance metrics

• Use included explainability tools (XAI) and batch analysis for further insights.

---

7. To Run Site

• Go to Web_Project Folder

cd Web_Project

• Run the Streamlit code

streamlit run final.py

---

Contributors

1. Paranthagan S
2. Nandana M