[Time series analysis of abnormal heart beat signals]

The variability of electric and magnetic signals from the heart during the depolarization phase is investigated. A signal processing method is developed which provides estimates for the beat-to-beat variability of the QRS-complex. The method is based on the decomposition of the depolarization signal into bandpass signals by means of the Morlet wavelet transform. The beat variability of the depolarization signal is estimated by normalized variances of the envelope and instantaneous frequency of bandpass signals. Time intervals of the bandpass filtered depolarization signals having a high signal-to-noise ratio are selected by applying an analysis based on phase statistics. The method was tested by experimental data taken from ECG and MCG measurements of healthy persons and patients prone to malignant ventricular tachycardia (VT) or ventricular fibrillation (VF). Results suggest that the calculated variance parameters permit the characterization of beat variable depolarization signals and distinguish VT/VF patients from healthy persons. The method developed can be used to obtain additional information concerning abnormal heart signals which is attenuated when applying signal averaging.