06 / 14 — Computer Vision & Signal Processing
ECG Signal Feature Extraction & Classification
Signal-processing pipeline for extracting and classifying features from ECG data in real time.
ecg-signal-feature-extraction-classification.build
Overview
What it does
Signal-processing pipeline for extracting and classifying features from ECG data in real time.
The Problem
Why it needed building
Raw ECG waveforms are noisy, and clinically useful features like R-peaks need to be extracted reliably enough to trust downstream classification.
The Approach
How it was engineered
Built a SciPy/NumPy signal-processing pipeline to isolate and extract R-peak and waveform features from 12-lead ECG data, with scikit-learn handling classification on top.
The Outcome
Where it landed
The pipeline hits 99.1% R-peak detection accuracy on the MIT-BIH benchmark dataset while processing all 12 leads in real time at 1,000 samples/sec.