ECG Processing and Feature Extraction

This package performs ECG processing on custom datasets followed by feature extraction.

1. R-Peak Detection - Detect R-peaks in 12-lead ecgs using Neurokit2 tool.
2. ECG Delineation - to detect P/QRS/T onset/peak/offset.
3. Feature Extraction - Compute baseline features such as heart rate and ecg-related intervals (PR interval, QT interval, etc.)

Workflow

• Load ECG data from a specified dataset.
• Perform cleaning and preprocessing of ECG signals.
• Detect R-peaks and delineate ECG signals into their physiological components.
• Extract features from the cleaned ECG data.
• Save the extracted features and annotations to output files.

Installation

1. Clone the repository:

   [email protected]:dieterich-lab/ecg_processing.git
   cd ecg_processing
   

2. Install the package:

   pip install -r requirements.txt
   

3. To run the script:

   python main.py
   

Configuration

Edit the config.json file to specify dataset paths and output directories.

Data Loader and Formats

Currently, the pipeline is tested with 3 different datasets i.e. Mediconnect, UK Biobank and MIMIC-IV ECGs. The data loader script for each of these datasets is available under data_loader/ directory.

Now, to use your own dataset, create a data loader under the same directory,

Expected Output

1. A NumPy array of ECG signals (signal_array) with shape (n_samples, n_leads, n_timepoints).
2. The sample frequency (samp_freq).
3. A predefined list of channel sequences (CHANNELS_SEQ).