Research On The Topological Properties Of Alzheimer's Disease Dynamic Brain Network Based On Graph Theory

Alzheimer's disease(AD)is a neurodegenerative disease,the cause of which is not yet clear,and it is a worldwide problem.The combination of complex networks and graph theory methods makes it possible to analyze from the whole brain.Due to the complexity of the brain,the quantification and measurement of the brain are still preliminary.Aiming at the key problem of the detection of AD abnormal brain networks and brain regions,this article starts from the dynamic functional brain networks of AD patients and healthy controls to study the dynamic changes of brain networks.The main innovations are as follows:First,an algorithm for detecting abnormal brain networks based on local efficiency is proposed.First,the group information guided ICA algorithm is used to identify spatial independent components and determine the brain network;second,the sliding window technology and k-means algorithm are used to analyze the clustering state of the dynamic brain network;Finally,through the abnormality of the local efficiency of the dynamic brain network in each state,find the functional brain network with abnormality.The results of the study show that for the AD group and the healthy control group,33 independent components and 30 independent components were obtained using the GIG-ICA algorithm,and finally 12 functional brain networks with physiological significance were established,including Sensorimotor Network,Posterior Salience Network,the Default Mode Network,Right Executive Control Network,Basal Ganglia Network,Higher Visual Network,Primary Visual Network,Visuospatial Network,etc.The k-means clustering algorithm was used to obtain six clustering states.The clustering states of the two groups of data changed significantly.Through the two-sample t test,Brain networks with abnormal local efficiency and node degree include Sensorimotor Network,Posterior Salience Network,Executive Control Network,Default Mode Network.These networks are related to human emotion regulation and cognitive function,and to some symptoms of AD patients.Among them,the default mode network plays an important role.Second,a Spearman abnormal brain region detection algorithm based on nonlinear correlation is proposed.First,the Spearman and sliding window methods are used to construct a dynamic functional brain network for all the subjects'data;second,the topological properties of the dynamic functional brain network of AD patients and healthy controls are calculated;finally,the abnormal brain is found through the difference of the topological properties Area to provide new insights into the clinical diagnosis of AD.The results of the study show that for different sparse thresholds,the dynamic functional brain network in the AD patient group and the healthy control group has small world attributes,that is?(29)1,but the small-world attribute of the Alzheimer's disease group changed significantly.The shortest path length(L)of the Alzheimer's disease group was higher than that of the healthy control group,and the clustering coefficient(C_p)was lower than that of the healthy control group,indicating that the speed of information exchange between brain regions of Alzheimer's disease patients was lower than that of normal people,which may cause cognitive changes in Alzheimer's disease patients.We calculated two groups of data dynamic functional brain network node attribute node betweenness,by double samples t test,finally established the abnormal areas of the brain,Precentral gyrus(PreCG),Superior frontal gyrus,orbital(SFGmorb),Rolandic operculum(ROL),Insula(INS),Hippocampus(HIP)and Parahippocampal gyrus(PHIP),Superior occipital gyrus(SOG),Fusiform(FG),Posterior cingulate(PCC)and Lentiicular nucleus,pallidum(PAL),These brain regions are mainly located in the default mode network,which is related to clinical symptoms such as memory loss,cognitive decline,and inability to take care of themselves in patients with Alzheimer's disease.These results validate the results of this article and provide new insights for the study of AD brain networks.