Study On EEG Peceptual Pattern From Local To Global

Brain is the most complex system on structure and function as we known at present,and electroencephalogram(EEG)becomes the efficient research tool of brain activity with the advantages of non-invasive,convenience and low-cost.The local and global perceptual pattern of brain was studied with EEG.A single-channel EEG pattern recognition algorithm on basis of ensemble empirical mode decomposition and recurrence quantification analysis is proposed for analysing local perceptual pattern of brain.With the application of this algorithm on studying single-channel EEG pattern,hypoxic condition of brain could be detected at early stage.Meanwhile,the variation of degree of hypoxia could be described by EEG pattern.As various regions of brain may need coorperation with sensory stimulation or process of cognitive activity,information in local area cannot represent the whole perceptual pattern.Therefore,the study is extended from single channel to multiple channels.It is necessary that common information in the pattern be extracted from different regions when analysing global pattern.Therefore,the multivariate extension of empirical mode decomposition was applied on studying global EEG pattern.A multi-channel EEG pattern recognition algorithm based on multivariate empirical mode decomposition(MEMD)is proposed in the paper.The common pattern could be extracted by MEMD,and then its sychronization is analysed,by which the variation of global brain functional connectivity in hypoxic condition could be described much clearer.Using a high-density EEG array highly improves the spatial resolution of EEG when the EEG spatiotemporal patterns of painful stimulation are researched.The application of MEMD aviods tedious optimizing on band of carrier wave and reduces effect of individual difference of subjects as well.The improved algorithm of feature extraction is more suitable for scalp EEG,and could distinguish the EEG spatiotemporal patterns of stimulation on different fingers.