Detection of atrial fibrillation from the surface electrocardiogram using magnitude-squared coherence

Automated arrhythmia interpretation systems suffer in their ability to detect atrial fibrillation from surface electrocardiograms (ECG). This study examines the feasibility of detecting atrial fibrillation from the ECG using magnitude-squared coherence (MSG). Surface leads II and V1 were evaluated for mean MSG, R-R variability and percent power during atrial fibrillation, atrial flutter and sinus rhythm. Results show that mean MSC and R-R variability discriminate atrial fibrillation from atrial flutter and sinus rhythm. Most striking was the ability of mean MSC to differentiate atrial fibrillation from atrial flutter