Multiscale Community Detection In Functional Brain Network Using Dynamic Time Wrapping

The brain is a complex and magical organization system,which can process the received external information and promote human to complete the corresponding cognitive process.Knowing the relationship between brain structure and executive function is the key to understanding human thinking process and exploring brain diseases.In recent years,applying graph theory into studying brain function and view iting as a complex network is very popular,which can vividly reveal the process of information transmission in functional brain network.In addition,community structure is one of the basic characteristics to describe complex network structure and the internal potential relationship of the network.Therefore,analyzing the relationship between brain regions through community detection is important to understand the structure of brain tissue and explore some brain diseases.In the traditional construction of brain function network,Pearson correlation and Partial correlation are often used to describe the correlation between brain nodes.They can only describe the synchronization of time series in brain nodes,But there are also causal effects between the time series of brain nodes.In addition,there may be slight differences in the number of brain communities and the density of brain network due to the differences between individual brains.Therefore,exploring the density and communities in individual brain network is very important.This paper uses Dynamic Time Wrapping(DTW)to construct brain function network and detect community in brain networks under different densities.In addition,new measures,the Variation of Information(VI)and the Normalized Mutual Information(NMI)are proposed along with modularity as criteria for examining the best community structure of the binary network and defining the network density.The experimental results show that the brain function network based on our method has more obvious "small world" effect and is closer to the real human brain network.This is not only beneficial for us to detect the communities in the brain network,but also to discover the fine-grained brain connected communities in the brain network,such as the relationship between cerebellum and brain,and Parietal Inferior network including angular gyrus and sup ramarginal gyrus.So this paper provides a new research method for the exploration of brain diseases and individual differences.Finally,this paper explains the rationality of these new findings from the perspective of neurobiology.