Effective Brain Network Analysis Of Schizophrenia Base On Granger Causality

Schizophrenia is a severe disease of the brain,which is characterized by the disintegration of thought processes and emotional reaction,clinical symptom complex,involving feeling,thinking,emotion and other aspects,affect human cognitive function,and performance are different between different individuals.Because of the various symptoms and combinations of schizophrenia,the cause of schizophrenia has been debated from the past to the present.Because of the differences in clinical manifestations and individuals of schizophrenia,it is of great practical significance to study the differences in brain mechanisms between patients with schizophrenia and normal people.In the research of brain network,functional brain network is in the leading position.However,functional brain network can only reflect whether there is a functional relationship between different brain regions,and cannot reflect the flow of information in the brain network,that is,functional brain network corresponds to undirected graph.In order to reflect the information flow in the brain network,the researchers developed the brain network based on the functional brain network.In essence,efn is a special functional brain network,which is the same as the functional brain network in terms of data preprocessing and node definition.The difference is that efn can reflect the information flow between different brain regions,making the undirected graph of functional brain network become directed graph.In the study of causality brain networks,Granger causality(GC)method has obvious advantages,it is not affected by prior knowledge,and more emphasis on time sequence,so the study of causality brain networks is mostly based on this method.Based on previous studies,this paper utilizes Partial Directed Coherence analysis(PDC)based on the granger causality model to build a specific efficient brain network with EEG signals,and then analyzes the topological structure of those networks.The main work of this paper is as follows:First,the EEG signals of 64 lead resting state were used to construct the brain network for the two groups of subjects(schizophrenic patients and normal people).Firstly,the prior knowledge of eeg were used to remove the interference and artifacts of the data.Second,with multiple variate autoregressive model(MVAR)with appropriate order.Then,the causality of each channel is quantified by using the partial directional coherence analysis method based on granger causality model,and the adjacency matrix of the brain network is generated.Finally,appropriate thresholds should be selected to binarize the quantized causality,so as to remove some unimportant connection edges and reduce the size of the network.Secondly,the topological structure of the factor-effect brain network of schizophrenic patients and normal people was analyzed,and the factor-effect brain network was analyzed from the aspects of global attribute and local attribute respectively.In terms of global properties,through the analysis on the two groups of subjects under different frequencies due to the effect of brain network average clustering coefficient,average path length and the network global efficiency shows: two groups of subjects on average clustering coefficient and average path length was higher than that of patients with schizophrenia,but there is no significant difference between two groups of participants,and on the global network efficiency,normal person has high global network efficiency,and between the two groups had no significant difference.On the local properties,separately analyzes the schizophrenic patients and normal person due to the effect degree of brain networks,degree distribution,betweenness,tightness,local efficiency,by degrees,the degree distribution,order number,tightness after the analysis of the relationship between the nodes,the patients with schizophrenia in the right hemisphere of the brain's frontal and temporal lobe in distribution,the core of the network node and normal person is mainly located in the left hemisphere of the brain's frontal lobe,temporal lobe and parietal lobe;By analyzing the local efficiency of the two groups of subjects,we found that the schizophrenic patients have higher local efficiency of the network.