AFib Detection System
Developed a MATLAB-based signal processing system to classify atrial fibrillation vs. sinus rhythm using frequency-domain ECG analysis.
Project Overview
As part of a team project in Systems and Signals Engineering, we developed an automated system to detect atrial fibrillation (AFib) in ECG signals using frequency domain analysis. AFib affects an estimated 12.1 million people in the U.S. and can cause serious complications including blood clots, stroke, and heart failure.
Technical Approach
Our system processes ECG signals through several key steps:
- Signal Processing: Applied high-pass filtering to isolate relevant frequency components
- Fourier Analysis: Used Fourier transforms to analyze frequency domain characteristics
- Feature Extraction: Focused on TP segments of ECG where AFib characteristics are most apparent
- Classification: Validated results using confusion matrix analysis
Key Findings
Our analysis successfully differentiated between AFib and normal sinus rhythm by identifying unique frequency domain characteristics. In AFib, the frequency within the TP segment showed significantly higher power (approximately 3-6 times greater) compared to normal sinus rhythm.
Technical Implementation
- Platform: MATLAB for signal processing and analysis
- Data: 6 trials of synthetic ECG data
- Methods: High-pass filtering, Fourier transforms, power calculations
- Validation: Confusion matrix analysis for classification accuracy
Project poster: "Frequency Analysis of ECG: Detecting Sinus vs. AFib Rhythm"
Clinical Impact
This system provides an objective method for AFib detection that could assist clinicians in early diagnosis. The automated frequency analysis reduces reliance on subjective interpretation and could be integrated into wearable ECG monitoring systems for continuous patient monitoring.
Presentation
We presented our findings to an audience of 40 people, proposing clinical integration strategies for early AFib detection in wearable ECG systems. The project demonstrates the potential for signal processing techniques to improve cardiovascular diagnostics.