Comparison of bivariate and contextual analysis in discrimination of atrial fibrillation from surface electrocardiogram

Automated arrhythmia interpretation systems fail to reliably detect atrial fibrillation from surface electrocardiograms (ECG). Recent investigations suggest that magnitude-squared coherence (MSC) may be a reliable discriminator of atrial fibrillation from non-fibrillatory atrial rhythms. This study compares MSC to traditional contextual analysis (explicit P wave detection) in the discrimination of atrial fibrillation from sinus rhythm and atrial flutter. Standard 12-lead ECGs acquired from patients exhibiting atrial fibrillation, atrial flutter and sinus rhythm were analyzed using both MSC and contextual analysis techniques. Results show that mean MSC discriminates atrial fibrillation from sinus rhythm and atrial flutter in surface ECG with significantly less overlap than that of contextual analysis. Unlike contextual analysis, MSC does not require explicit P wave detection and utilizes simpler detection criteria.