The Research Of Community Structure Difference Of Resting-state Functional Brain Network In Tristimania

Most complex systems,including social,economical and biological networks,have a potential modular structure.Modularity is thought to be one of the main organizing principles of complex systems.Graph theory provides a method for us to quantitatively analyze complex networks.The human brain can be abstracted into a complex network.Brain region of the human brain can be abstracted as a node.Functional association which exist in brain regions can be abstracted into an edge.Community structure is one of the basic properties of the complex network.Nature of the structure and function relationships of brain network can be understood profundly by analyzing the community natrue of the network.Here,making a Synchronizing analysis of low-frequency fluctuation signal of the fMRI,we build a resting state functional brain network and divided into several major modules based on the greedy algorithm of "heap structure"algorithms.Then,we use methods of graph theory(community structure and the topological role of nodes)to analyze resting state human brain functional network.Finally,we explore abnormality of depression patients in the nature and connectivity of brain networks.Through experimental analysis found that there are significant differences in terms of community structure and roles of nodes of brain regions of human brain functional network between patients with depression and normal human.These differences may be the underlying causes of the onset of depression.Firstly,we show that the modularity of both the normal subjects and Depression patients was significantly greater than random.Significant difference in modularity did not find in two groups through calculating AUC of modularity in the sparsity ranged from 2.5%to 10%,but find in some particular sparsity(5%,7%).Secondly,for normal group,depression group and comparable random networks,With the increase of the brain network sparsity,Modularity decreases monotonically.In the threshold space ranges,modularity is a decreasing function of the number of edges.When the network have 100 edges,the modularity is the highest.Thirdly,we introduce the concept of "small-nuclei" and "intimacy" to analyze Depression patients and find that compared to the normal subjects,the number of inter-modular connection of the Depression patients changed.Fourthly,the experiments show that the differences of modules composition,the node role of the brain network and the role of modules exist in two groups.The discovery of these differences provides some important research and clinical value for us in disease diagnosis.Finally,we select some attribute values containing global-attribute arnd node-attribute from the modularity Q as well as Z-values and P-values of 90 nodes,which the selection criteria is Classification results better.Then a classifier was constructed through training on these properties.Here we screened the top 10 attributes from the atrributes of the better classification performance,and selected 70 percent of the sample to train these properities in order to construct a classifier.Finally,the classifier was tested with the remaining 30 percent of the sample.The results show that the correct classification rate is up to over 90%.