Self-supervised pre-training for ECG representation with inspiration from recent advancements in transformers in Natural Language Processing and Computer Vision.
| Name | # records |
|---|---|
| St Petersburg INCART 12-lead Arrhythmia Database | 75 |
| PTB Diagnostic ECG Database | 549 |
| PTB-XL, a large publicly available electrocardiography dataset | 21,837 |
| China Physiological Signal Challenge 2018 | 6,877 |
| CSPC extra/unused dataset | 3,453 |
| Georgia 12-lead ECG Challenge (G12EC) Database | 10,344 |
| A 12-lead electrocardiogram database for arrhythmia research covering more than 10,000 patients | 10,646 |
| Test set from paper Automatic diagnosis of the 12-lead ECG using a deep neural network | 827 |
| MIT-BIH Arrhythmia Database | 48 |
| MIMIC-IV-ECG: Diagnostic Electrocardiogram Matched Subset | 800,000 |
| Abdominal and Direct Fetal ECG Database | 5 |
Note that all entires apart from Test set from paper Automatic diagnosis of the 12-lead ECG using a deep neural network are part of the PhysioNet - Computing in Cardiology Challenge 2021 (CinC21). We collect the dataset from the original publishing source if available since the versions from CinC21 had records removed.
1< Have the datasets linked above downloaded.
2> Modify the DIR_DSET variable in file data_path.py
as instructed.
A folder named as DIR_DSET should be kept at the same level as
this repository, with dataset folder names specified as
in config.json.