Skip to content

paperchae/HeartBox

Repository files navigation

HeartBox

Description

  • This project is to detect heart disease using biomedical signals such as ECG, PPG and ABP through Pytorch.
  • For Data,
    • Data Inspection
    • Preprocessing
    • Feature Extraction
    • Augmentation is provided
  • Training and Evaluation is provided
  • This project will tie together all the functionality in MIMIC and CNIBP, so that it can be used as a single package for heart disease detection.

Preprocessing

preprocess

  • Baseline Correction
  • Removes baseline wandering created by respiration, body movement, etc.
  • Denoising
  • Removes noise created by power line, muscle contraction, etc.

Data Augmentation

Augmentation


Datasets

Vital Signal Datasets

MIMIC-IV-ECG

  • 12 leads ECG, 10 seconds length
  • Patients: approximately 160,000
  • Records: 800,000

MIMIC-IV Waveform Database

  • Contains ECG, ABP, PPG, Resp, etc.
  • Patients: 198
  • Records: 200 (10,000 records upcomming)

MIMIC-III Waveform Database

  • Contains ECG, ABP, PPG, Resp, etc.
  • Patients: approximately 30,000
  • Records: 67,830

PTB Diagnostic ECG Database

  • TBD

Clinical Database

MIMIC-IV

  • Patients

Results

Results

mindmeister

Contact

JONGEUI CHAE : [email protected]

Please feel free to contact, open for any collaboration.

Releases

No releases published

Packages

No packages published

Languages