ECG Heartbeat Classification via Deep Neural Networks

This project implements deep neural network models to classify heartbeats from electrocardiogram (ECG) signals into different categories of arrhythmia. The dataset combines signals from two well-known sources:

• MIT-BIH Arrhythmia Dataset
• PTB Diagnostic ECG Database

The goal is to accurately identify abnormal heart rhythms and support the development of computer-aided diagnosis systems in cardiology.

🎯 Project Goal

• Train a deep learning model to categorize ECG heartbeat signals.
• Detect and classify arrhythmia types from ECG recordings.
• Explore transfer learning for improving classification accuracy.
• Provide insights into the most distinguishing features of ECG waveforms.

📂 Dataset Overview

1. MIT-BIH Arrhythmia Dataset

• 48 half-hour excerpts of two-channel ECG recordings
• 47 subjects (recorded between 1975–1979)
• 109,446 heartbeat samples classified into 5 categories:
  • N – Normal beats
  • S – Supraventricular ectopic beats
  • V – Ventricular ectopic beats
  • F – Fusion beats
  • Q – Unknown beats