Temporal evolution for the phase histogram of ECG during human ventricular fibrillation

A novel approach to momentary/instantaneous morphological assessment of phase histograms, extending phase statistics analysis, is used to investigate electrocardiograms during ventricular fibrillation (VF) in humans. By using empirical mode decomposition (EMD) and the Hilbert transform, we calculate the instantaneous phase of intrinsic mode functions (IMFs) in Holter data from 16 individuals, and construct the corresponding momentary phase histograms, enabling us to inspect the evolution of the waveform of the time series. A measure defined as the difference between the integrals of the probability distribution density of phase in different regions is then used to characterize the morphology of the momentary histograms and their temporal evolution. We find that the measure of morphology difference allows near perfect classification of the VF data into survivor and non-survivor groups. The technique offers a new possibility to improve the effectiveness of intervention in defibrillation treatment and limit the negative side effects of unnecessary interventions. The approach can be implemented in real time and should provide a useful method for early evaluation of (fatal) VF.