MSC AND PHASE SYNCHRONIZATION ANALYSIS FOR CLASSIFICATION OF EEG

Computer Engineering

EEG signals are often used for disease diagnosis and behavioral analyses. These signals are highly non-stationary. Therefore a common practice of EEG analysis is, breaking it into many band limited components and then extrapolating the underlying features of these temporal components. This paper is about analysis of mean spectral coherence and phase synchronization of EEG components, across various sub-bands. Coherence is studied for different pair of EEG signals, acquired from corresponding locations of the two hemisphere of brains. Coherence feature is unrestricted to amplitude distortions which are caused by various signal processing steps. This work is centered on MSC computation by using wavelet transform. By using suitable frequency resolutions, or in other ward by using different number of samples while computing FFT for different wavelet sub-band components, wavelet based computation results into time-frequency resolved MSC coefficients. This helps in preserving the underlying less powerful neuronal dynamics, at some frequencies, which would otherwise disappeared due to presence of more powerful harmonically related frequencies. The results show that similar synchronization or association exist between MSC peals and the corresponding phase across the sub-bands, however this association varies in specific ways depending on external stimulus. For this experimentation external stimulus is considered through five different tasks performed by subjects during EEG acquisition. The variations are better observable in theta, beta and gamma sub-bands.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.