Coherence analysis of multichannel epicardial electrograms for quantification of organization during ventricular fibrillation

A new technique to characterize the organization of electrical activity in the myocardial muscles is presented. It employs magnitude-squared coherence (MSC) spectrums between unipolar electrograms recorded from the epicardial surface. In a canine study, unipolar electrograms were recorded (sampling rate 1,024 Hz) from 112 epicardial sites during both normal sinus rhythm (NSR) and ventricular fibrillation (VF). MSC-spectrums with respect to the electrogram from a reference site were computed using the weighted (Hanning window) overlapped (75% overlap) segmentation (segment length 512; 5 segments) fast Fourier transform approach. Average coherence at dominant frequencies during NSR was found to be 0.92±0.01, and during VF 0.41±0.03 (p<10^-49). Further, the averaged MSC between electrograms from two sites was found to decrease with increasing distance during both NSR and VF. Also, the dispersion of MSC was found to be greater during VF. It is concluded that coherence analysis of epicardial electrograms could be employed to characterize the myocardial electrical activity during different rhythms, and quantify their levels of organization