Time-frequency analysis of ventricular late potentials.

The time-frequency characteristics are studied of averaged and filtered ECG records from 21 patients with sustained ventricular tachycardia and 29 healthy control subjects. Simulated data as well as real ECG records reveal the detection accuracy of the wavelet transform of signals with late potentials. The wavelet-transforms of preprocessed ECG signals are plotted in the time-frequency plane. These representations of the signals are well suited to describe the different characteristics of the patients and healthy subjects. A quantitative discrimination was performed with a sensitivity of 90% and a specificity of 72% by the energy underneath the squared modulus of the time-frequency distribution plots of the computed wavelet transforms.