Wavelet Analysis of High-Resolution Signal-Averaged Electrocardiograms in Postinfarction Patients with Bundle Branch Block

Abstract

Time-domain analysis of the signal-averaged electrocardiogram (SAECG) can accurately predict risk of sustained ventricular tachycardia (VT) in patients with previous myocardial infarction (MI). Unfortunately, these patients often have bundle branch block (BBB) that obscures late potentials. We hypothesized that wavelet analysis might help predict VT risk in patients with BBB. We identified subjects with coronary disease and BBB who had undergone SAECG and programmed ventricular stimulation (PVS). We applied a modulated Gaussian wavelet to transform the ECG signal and looked for singularities in the wavelet coefficients. SAECG and PVS were obtained in 32 patients. Half had inducible sustained monomorphic VT by PVS and half had no inducible VT. There were no significant clinical differences between the groups. Comparing the number of singularities, we found no significant difference between the groups. Compared to previous work in patients without BBB, our patients with BBB had on average four times more singula- rities using an identical analysis technique. In the presence of BBB, abnormal myocardial activation patterns can generate ECG waveforms with time–frequency characteristics similar to those of cardiac late potentials. Wavelet-based methodologies may have limited ability to distinguish late potentials from the disordered ventricular activation occurring with BBB alone.