Heartbeat Classification based Diagnosis

This is a general prototype for machine learning based heartbeat classification. Label each heartbeat in ECG. R peak detection is first performed to locate each heartbeat. Each heartbeat then will be classified by a CNN. Corresponding label will be printed on the top of each R peak. The CNN is trained using a few throusand labeled beats. Abbreviations is listed below:
Normal: normal, AAP: Aberrated Atrial Premature, AP: Atrial premature, VE: Ventricular Escape, NE: Nodal escape, NP: Nodal Premature, LB: Left Bundle Branch Block, RB: Right Bundle Branch Block.

Dependencies

• Tensorflow
• Numpy
• Matplotlib

  run example

  ~$ git clone https://github.com/KChen89/Heartbeat-Classification-based-Diagnosis.git
  ~$ cd /your folder
  ~$ python3 hb_classifier.py ecg.dat
  ~$ python3 hb_classifier.py ECG_sample.dat

  The model is trained using ECG sampled at 360 Hz, and length of each beat truncked are pre-defined.

More

• [x] Beat classification.
• [x] Print labels on top of R peak.
• [x] Combining R peak detection.
• [ ] Include more type of beat.
• [ ] Other platform (Mobile).

Reference

[1] K. Chen, L.S. Powers, J.M.Roveda, "Noise-Invariant Components Analysis for Wearable Sensor based Electrocardiogram Monitoring System".