Diagnostic implications of spectral and temporal analysis of the entire cardiac cycle in patients with ventricular tachycardia

Abstract

Background. Available methods for analysis of signal-averaged electrocardiograms (ECGs) have a low-positive predictive accuracy for identifying patients at risk for ventricular arrhythmias. Identification of the spectral and temporal features in ECGs that distinguish patients prone to ventricular tachycardia (VT) is a prerequisite to implementing refinements in methods that increase the diagnostic power of the signal-averaged ECG. Methods and Results. Fast Fourier transforms and time-domain reconstructions based on inverse fast Fourier transforms were computed over the entire cardiac cycle of signal-averaged ECGs of sinus beats from 40 patients with myocardial infarction and sustained VT, 41 with infarction without VT, and 20 normal controls. Ventricular depolarization and repolarization were analyzed by procedures that obviate limited resolution due to short data segments and window functions. Spectral magnitudes of ECGs from patients in each group were compared, and the phase data were used for time-domain reconstructions to determine the temporal distributions of distinguishing frequency bands during the cardiac cycle. Magnitudes of 1-7-Hz frequencies were increased (from p < 0.05 to p < 0.00001), and magnitudes of 13-56-Hz and 70-128-Hz frequencies were decreased (from p < 0.05 to p < 0.00001) in the spectra of ECGs from patients with VT compared with patients without VT. Time-domain reconstructions demonstrated that 1-7-Hz frequencies were detectable throughout the QRS complex, ST segment, and T wave in ECGs from each group. The 13-56-Hz and 70-128-Hz frequency bands not only contributed to the terminal QRS and ST segment but were also detectable throughout the QRS complex of ECGs from patients with VT. Conclusions. Results define new spectral and temporal features in signal-averaged ECGs from patients with VT that are excluded from analysis by available techniques that limit the bandwidth or restrict interrogation to portions of the cardiac cycle. These findings provide an objective basis for developing new indexes for signal-averaged ECG analysis.