Analysis of Schizophrenic EEG Synchrony using Empirical Mode Decomposition

Phase synchronization of neuron oscillations is believed to be abnormal in schizophrenic patients, and has attracted the interests of many researchers. For multicomponent signals such as electroencephalogram (EEG), since the trajectory rotates around multi-centers in the complex plane, phase is not meaningful. Usually Fourier based (linear) time-frequency decomposition (e.g. wavelet transform) has been used to analyze the degree of phase synchronization in each time-frequency bin. Considering the nonlinear nature of neuron oscillations, in this paper, we analyze phase synchronization from a different view point by decomposing the signal into several intrinsic modes using empirical mode decomposition (EMD), each of which can be nonlinear and non-stationary. The results obtained from simulated signals and real EEG signals indicate that EMD might be more accurate to the real dynamics of the oscillation process than time-frequency decomposition.