Score-based-ECG-Denoising

This repository contains the codes for DeScoD-ECG: Deep Score-Based Diffusion Model for ECG Baseline Wander and Noise Removal

The deep learning models were implemented using PyTorch.

Dataset

bash ./data/download_data.sh

Train the model

!python -W ignore main_exp.py --n_type=1

!python -W ignore main_exp.py --n_type=2

Evaluation the model

!python -W ignore eval_new.py

Citing our work

H. Li, G. Ditzler, J. Roveda and A. Li, "DeScoD-ECG: Deep Score-Based Diffusion Model for ECG Baseline Wander and Noise Removal," in IEEE Journal of Biomedical and Health Informatics, doi: 10.1109/JBHI.2023.3237712.

Acknowledgement

The data preprocessing is directly taken from DeepFilter.