Session SA4.2

Modified Wavelet Bicoherence as a Diagnostic Tool for Very High Frequency Peaks in Cardiovascular Signals of Heart Transplant Patients

K Keissar*, O Gilad, S Akselrod

Tel Aviv University
Tel Aviv, Israel

Wavelet bicoherence (WB) has been introduced recently as a technique for revealing non linear modulations among interacting noisy oscillators. WB has been shown to have advantages over the traditional Fourier-based bicoherence estimates and to be useful in the analysis of multivariate dynamic signals of the cardiovascular system. We propose a modified estimator for WB which has several advantages over the current estimator. Our modified wavelet bicoherence (MWB) is based on frequency shift of all the analyzed signals using Hilbert transform and a mixer, which enables an optimal time and frequency resolution corresponding with the wavelet transform (WT). In addition, the MWB estimator could enable bicoherence analysis in the absence of one of the two source signals. This could be useful for cases when only partial knowledge of the system in question is available. First, we have reviewed and developed the mathematical formulation of the MWB. Second, the statistics and error estimation were presented empirically using simulated data. Limitations of the method and threshold criterion were established. Finally, the MWB was used as an analysis tool for the investigation of observed peaks in the very high frequency (VHF) band in heart rate (HR), blood pressure and respiration signals of 17 (age: 53± 11 years) heart transplant patients and 16 (age: 41± 7 years) normal subjects. Applying the MWB in conjunction with WT over the two groups, provided a novel presentation of the bicoherence dynamics which allowed linking the VHF peaks to their originating source in about 55% of heart transplant patients and 30% of normal patients. In conclusion, The MWB estimator is a promising tool for detection and estimation of dynamic and transient phase coupled nonlinear interactions characterizing the cardiovascular system.

(Abstract Control Number: 137)