Dickhaus H, Khadra L, Brachmann J
Med. Informatik, Universität Heidelberg/Fachhochschule Heilbronn, Germany.
Comput Methods Programs Biomed. 1994 Jun;43(3-4):185-92. doi: 10.1016/0169-2607(94)90069-8.
Late potentials of preprocessed ECG recordings from patients with sustained ventricular tachycardia and healthy control subjects are investigated by the wavelet transformation. The advantages of this non-stationary signal processing method are discussed in comparison with FFT spectrograms of artificial test signals. The energy distribution plots of these wavelet transformed test signals demonstrate the superiority in detection accuracy and frequency resolution compared with the corresponding spectrograms. For this reason a quantitative discrimination between the two clinical groups of patients and healthy subjects was performed by calculating the scalograms of the wavelet transformed ECG signals in the time-frequency plane. The energy in the frequency band of 100-300 Hz during the last time segment of the QRS complex yielded best classification results.
