Large-scale Granger Causal Brain Network Based On Resting-state FMRI Data

The human brain is a continuously communicating dynamic network,it has long been recognized that the interaction between the brain neurons is directional.Being able to assess the directionality of brain neuron interactions is an important ability to understand neural network interactions in the brain,and to study the direction of information flow between brain regions can highlight the role of each region in the network.The causal connection among small-scale regions based on resting-state fMRI data has been extensively studied and a lot of achievements have been demonstrated.However,the causal connection among large-scale regions has not been well elucidated yet.How the brain regions causally influence each other is still largely unknown.In this study,Granger causality analysis method and graph theory method were combined to conduct a preliminary study on large-scale brain networks,the main work of this research includes:(1)The fMRI data of 103 healthy subjects were preprocessed by SPM,DPSARF and other software,such as time layer correction,head motion correction and image space standardization processing,etc.,to obtain the time series of 90 brain regions.(2)We applied the global Granger causality analysis to a resting-state fMRI dataset for calculating causality between time series of 90 brainregions.The whole-brain was divided into 90 distinct regions by using the automated anatomical labeling(AAL)atlas.Then T test was performed on the connection matrix,and only values within the range of P<0.05 were retained.The results show that of the 8100 total possible causal connections,817 directed edges were identified to be significant.The analysis proves that the result is effective and reasonable.(2)The causal connectivity of four large-scale cognitive networks,the central executive network(CEN),default mode network(DMN),the dorsal attention network(DAN)and salience network(SN)have been investigated.The results show that,in CEN,the DlPFC,mDlPFC,DmPFC had obvious causal effect on the SPC.In DMN,The MPFC showed significant direct causal influences to the PCC.In DAN,the SPL may received both direct and indirect causal influence of the FEF,then from SPL to IPS,the IPS had significant causal effect on the FEF.In SN,the dACC imposed significant causal influences on the FIC.(3)The causal network is further analyzed by combining with the graph theory,the seven driving hubs and ten driven hubs of the whole brain,as well as the drivinghubs and driven hubs of four large-scale cognitive networks are obtained,the important regions of the brain network were determined.