AF Classification Backend

This repository contains backend code for ECG signal processing and atrial fibrillation (AF) classification, including signal quality assessment, noise detection, P wave and R peak detection, and AF analysis. It supports the PTB-XL dataset and simulated ECG data.

---

Table of Contents

• Features
• Project Structure
• Installation
• Usage
  • ECG Record Loading
  • Signal Quality Assessment
  • P Wave and R Peak Detection
  • Atrial Fibrillation Detection
  • Simulating ECG Data
• Configuration
• Testing
• Dependencies
• Troubleshooting
• License
• Acknowledgements

---

Features

• ECG Record Loading: Load and parse ECG records from PTB-XL or WFDB-compatible files.
• Signal Quality Assessment: Comprehensive checks for noise, flatline, saturation, and overall signal usability.
• P Wave and R Peak Detection: Robust detection algorithms for P waves and R peaks.
• Atrial Fibrillation Detection: Feature extraction and AF classification using RR intervals, P wave presence, Baseline and QRS width.
• Visualization: Plot ECG signals with detected features.
• ECG Simulation: Generate synthetic ECG signals (including AF) using NeuroKit2.
• Extensible: Modular code for easy integration and extension.

---

Project Structure

af_classification/
│
├── main.py                # Main entry point and workflow
├── p_wave.py              # P wave detection logic
├── af_detector.py         # AF detection and analysis
├── simulate.py            # ECG simulation using NeuroKit2
├── requirements.txt       # Python dependencies
├── .gitignore             # Git ignore rules
├── training2017/          # (Optional) ECG data directory
└── ... (other files)

---

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/af_classification.git
   cd af_classification

2. Create a virtual environment (recommended):

   python3 -m venv venv
   source venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

   If you use NeuroKit2 for simulation, also ensure:

   pip install neurokit2 PyWavelets

---

Usage

ECG Record Loading

Load an ECG record (e.g., from PTB-XL):

from main import Plotter

plotter = Plotter(data_path='./training2017')
record = plotter.get_record('A00001')
print("ECG shape:", record['signal'].shape)

Signal Quality Assessment

quality = plotter.comprehensive_quality_check(record['signal'][:, 0], record['fs'])
print("Signal quality:", quality['overall_quality'])

P Wave and R Peak Detection

p_wave_results = plotter.detect_p_waves_in_record(record)
plotter.plot_p_wave_detection(record['signal'][:, 0], p_wave_results, record['fs'])

Atrial Fibrillation Detection

from af_detector import AtrialFibrillationDetector

af_detector = AtrialFibrillationDetector(fs=record['fs'])
af_results = af_detector.detect_af_in_record(record)
print("AF Detected:", af_results['af_detected'])
af_detector.plot_af_analysis(record['signal'][:, 0], af_results)

Simulating ECG Data

import neurokit2 as nk
import wfdb
import numpy as np

fs = 300
ecg = nk.ecg_simulate(duration=10, sampling_rate=fs, heart_rate=90, noise=0.01)
ecg_signal = np.array(ecg).reshape(-1, 1)
wfdb.wrsamp('sim_sinus', fs=fs, units=['mV'], sig_name=['I'], p_signal=ecg_signal)

---

Configuration

• Data Path: Set the data_path argument in Plotter to your ECG data directory.
• Lead Selection: By default, lead I ([:, 0]) is used. Change the index for other leads.

---

Testing

You can add unit tests using pytest or your preferred framework. Example:

pytest tests/

---

Dependencies

• Python 3.7+
• numpy
• scipy
• matplotlib
• wfdb
• neurokit2 (for simulation)
• PyWavelets (for neurokit2)

Install all with:

pip install -r requirements.txt

---

Troubleshooting

• ModuleNotFoundError: Install missing packages with pip install <package>.
• ValueError: The length of the input vector x must be greater than padlen: Ensure your ECG signal is long enough (typically >30 samples for filtering).
• WFDB plotting errors: Pass only WFDB record objects to plot_ecg, not numpy arrays or dicts.

---

License

This project is licensed under the MIT License.

---

Acknowledgements

• WFDB Python
• NeuroKit2
• PTB-XL ECG Dataset ([PhysioNet](https://physionet.org/content/ptb