Application Of Time-varying Network In EEG Data Analysis

Many previous studies have shown that the human cognitive processes(i.e.,motor imagery)and brain diseases(i.e.,schizophrenia)are correlated with the interactions of multiple brain regions/networks.The network analysis was widely used to explore the human brain.Studying the patter of networks is helpful to understand the underlying neural mechanisms of brain at the diverse brain states and different scale levels.Thus,in this thesis,based on the high-resolution EEG data,we applied the adaptive directed transfer function(ADTF)method to the healthy subjects and schizophrenia patients to investigate the patters of time-varying network.In this paper,we have the below two main contributions:Firstly,we explored the patterns of time-varying network of the left-hand motor imagery(MI)and right-hand MI by using the ADTF method,respectively.Based on the MI-EEG data,we first extracted the time courses of ERD and ERS.We then constructed the time-varying networks using the ADTF for each subject.Our results found that 1)in the MI preparation stage,the networks showed symmetric connectivity patterns;2)in the ERD stage,the networks showed hemispheric lateralization connectivity patterns during the left-hand MI and right-hand MI;3)in the ERS stage,the network recovered a symmetric connectivity pattern.Moreover,we calculated the network properties for each time-varying network.Similar to the network topologies,the three MI stages also appear to be characterized by different network measurements.These findings facilitate to understand the neural mechanisms of MI.Secondly,we explored the time-varying network patterns of sensory gating of the healthy subjects and schizophrenia patients.The P50 could be recorded by the conditioning stimulus(S1)and testing stimulus(S2)by using auditory double clicks paradigm.The suppression ration(S2/S1)was more than 50% in patients with the schizophrenia.In the current study,we first extracted the P50 of the S1 and S2 stimulus for all subjects.Then,we constructed the time-varying networks by using the ADTF method.Finally,we compared the network patterns of S1 and S2 of healthy subjects,schizophrenia patients and both.We found that 1)the right temporal lobe could be observed the significant decreasing of brain activity to the second stimuli for the healthy controls,2)for the patients with schizophrenia,there did not exist the significant differences between the time-varying networks of the S1 and S2;and 3)comparing the patients with schizophrenia,the healthy controls showed the stronger information couplings from the right parietal lobe to the frontal lobe.We believe these findings could deepen our understanding about the neural mechanism of sensory gating,and could provide the fundamental basis for the clinical therapies of the schizophrenia.Finally,our findings consistently suggest that the time-varying network analysis was helpful for investigating brain cognition processes.The activity of the brain is a process of dynamic change.The time-varying network analysis can capture the dynamic change of the brain,facilitating to decode the neural mechanisms of cognition process and finding new biomarker for diagnosing brain diseases.