https://github.com/msadeqsirjani/ecg-ml
A Python tool for organizing the PTB-XL ECG dataset into a structured hierarchy of binary (normal/abnormal), superclass (NORM, MI, STTC, CD, HYP), and subclass (diagnostic codes) categories. Supports both 100Hz and 500Hz signals with automated file organization and detailed statistics generation.
https://github.com/msadeqsirjani/ecg-ml
data-preprocessing deep-learning ecg health-care python signal-processing
Last synced: 11 days ago
JSON representation
A Python tool for organizing the PTB-XL ECG dataset into a structured hierarchy of binary (normal/abnormal), superclass (NORM, MI, STTC, CD, HYP), and subclass (diagnostic codes) categories. Supports both 100Hz and 500Hz signals with automated file organization and detailed statistics generation.
- Host: GitHub
- URL: https://github.com/msadeqsirjani/ecg-ml
- Owner: msadeqsirjani
- License: mit
- Created: 2025-04-14T23:06:09.000Z (about 1 month ago)
- Default Branch: main
- Last Pushed: 2025-05-02T14:51:04.000Z (16 days ago)
- Last Synced: 2025-05-02T15:47:27.900Z (16 days ago)
- Topics: data-preprocessing, deep-learning, ecg, health-care, python, signal-processing
- Language: Python
- Homepage:
- Size: 79.8 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# ECG-ML: Missing Value Analysis and Classification
[](https://opensource.org/licenses/MIT)
[](https://www.python.org/downloads/)
[](https://www.tensorflow.org/)
[](https://wfdb.readthedocs.io/)
[](https://github.com/psf/black)
[](https://ecg-ml.readthedocs.io/en/latest/?badge=latest)
*Advanced ECG Signal Processing with Missing Value Analysis and Deep Learning Classification*
## 📊 Overview
This project implements a comprehensive pipeline for processing, analyzing, and classifying ECG signals. It includes functionality for handling missing values, preprocessing signals, and training deep learning models for ECG classification.
## 🏗️ Project Structure
```
.
├── data/ # Data directory (gitignored)
│ ├── raw/ # Raw ECG data
│ └── processed/ # Processed ECG data
├── logs/ # Training and processing logs (gitignored)
├── models/ # Trained models (gitignored)
├── results/ # Evaluation results (gitignored)
├── scripts/ # Shell scripts for setup and execution
├── src/ # Source code
│ ├── preprocess_ecg.py
│ ├── train_ecg.py
│ └── preprocess_missing_ecg.py
└── requirements.txt # Python dependencies
```
## ✨ Features
### 🔍 ECG Signal Processing
- Signal preprocessing and normalization
- Missing value handling and generation
- Multi-lead ECG support
- Configurable preprocessing parameters
### 🧠 Model Training
- Support for multiple classification types:
- Binary classification
- Superclass classification
- Subclass classification
- Configurable model architecture
- Comprehensive logging and monitoring
- Model size analysis and reporting
### 📂 Data Management
- Organized data directory structure
- Support for different lead configurations
- Efficient data loading and processing
## 🚀 Getting Started
### Prerequisites
- Python 3.x
- Git
- Virtual environment (recommended)
### Installation
1. Clone the repository:
```bash
git clone https://github.com/msadeqsirjani/ecg-ml.git
cd ecg-ml
```
2. Create and activate a virtual environment:
```bash
python -m venv env
source env/bin/activate # On Windows: env\Scripts\activate
```
3. Install dependencies:
```bash
pip install -r requirements.txt
```
## 📝 Usage
### 1️⃣ Data Download
```bash
./scripts/setup_ecg_data.sh
```
This script will:
- Create necessary data directories
- Download the ECG dataset from the source website
- Extract and organize the raw data files
### 2️⃣ Data Preprocessing
```bash
./scripts/setup_preprocess.sh
python src/preprocess_ecg.py
```
### 3️⃣ Missing Value Generation
```bash
./scripts/setup_missing_values.sh
python src/preprocess_missing_ecg.py
```
The missing value generation process:
- Takes 10-second ECG signal segments as input
- Introduces random zero-value noise segments
- Duration of noise segments varies from 1 to 9 seconds
- Multiple noise segments can be added to a single signal
- Preserves the original signal structure while simulating missing data scenarios
- Generates a new dataset with controlled missing value patterns
### 4️⃣ Model Training
```bash
./scripts/setup_training.sh
```
## ⚙️ Configuration
The project supports various configuration options:
| Category | Options |
|----------|---------|
| Lead Configurations | Different combinations of ECG leads |
| Classification Types | Binary, superclass, and subclass classifications |
| Model Parameters | Configurable through the training script |
| Preprocessing Parameters | Adjustable in the preprocessing script |
| Missing Value Parameters | Noise segment duration (1-9 seconds), segment count, distribution |
## 📊 Output
| Output Type | Location | Description |
|------------|----------|-------------|
| Processed Data | `data/processed/` | Preprocessed ECG signals |
| Trained Models | `models/` | Saved model checkpoints |
| Logs | `logs/` | Training and processing logs |
| Results | `results/` | Evaluation metrics and visualizations |
| Model Reports | `results/` | Size analysis and architecture details |
## 📚 Dependencies
- Python 3.x
- NumPy
- Pandas
- TensorFlow
- WFDB
- Scikit-learn
- Matplotlib
See `requirements.txt` for complete list of dependencies.
## 🤝 Contributing
We welcome contributions! Please follow these steps:
1. Fork the repository
2. Create a feature branch (`git checkout -b feature/AmazingFeature`)
3. Commit your changes (`git commit -m 'Add some AmazingFeature'`)
4. Push to the branch (`git push origin feature/AmazingFeature`)
5. Open a Pull Request
## 📄 License
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
## 📧 Contact
Your Name - [@msadeqsirjani](https://twitter.com/msadrqsirjani) - [email protected]
Project Link: [https://github.com/msadeqsirjani/ecg-ml](https://github.com/msadeqsirjani/ecg-ml)
---
Built with ❤️ by the ECG-ML Team