Research On Topological Properties Of AD Brain Network Based On Dynamic Functional Connectivity Analysis

Alzheimer's disease(AD)is a neurodegenerative disease that occurs mostly in old age,causing a decline in the patient's living ability,family and society economic pressure and psychological burden.The etiology is not clear,and the symptoms are relieved mainly by drugs that improve mental and cognitive.Based on resting state functional magnetic resonance imaging,this thesis proposes two methods for analyzing abnormal topological properties of AD dynamic functional brain network from the perspective of dynamic functional network.The main work of this thesis is as follows:First,an AD dynamic topology time-varying network analysis algorithm based on visible graphs is proposed.First,a dynamic functional connection network of each subject was constructed by combining time derivative multiplication and sliding window.Secondly,use graph theory method to calculate the time series of dynamic function network topological attributes.Finally,a local dynamic topology time-varying network and a whole brain dynamic topology timevarying network were constructed for each subject based on the visible graph algorithm.To explore the abnormal time characteristics of the dynamic functional network of patients with Alzheimer's disease from the perspective of local topology time-varying network and whole brain topology time-varying network.The results show that the local time-varying network of CN group changes regularly.With the aggravation of the disease,the degree of chaos in the dynamic fluctuation of functional connection is increasing.Compared with the CN group,significant differences in brain regions appeared in each stage of the AD course,and the degree of difference increased as the disease got worse.Local brain area abnormalities in AD patients mainly occur in the right intraorbital superior frontal gyrus,the cortex around the left talar fissure,the left middle occipital gyrus and the right hypooccipital gyrus.With the development of the disease,the shortest path length of the whole brain topology time-varying network decreases,and the small-world attribute increases.Our research results provide new insights into understanding the time characteristics of dynamic networks in the AD process from the abnormal characteristics of local topology timevarying networks and whole brain topology time-varying networks.Second,an AD dynamic brain network state analysis algorithm based on Louvain multilayer network modularization is proposed.First,combine the time derivative multiplication and sliding window to construct a dynamic functional network for each subject.Secondly,the multi-layer network modularity is embedded into Louvain's algorithm to obtain the multi-layer network maximization modularity and module division of AD patients and CN control group.Finally,the degree of modularity is used to divide the AD group and the CN group into different high and low modular periods,and the histograms of the z scores in the modules and the participation coefficients between the modules of each group are clustered to divide the integration and separation states.The results of the study showed that compared with the CN group,the specialized functional network connection pattern in the AD group was reduced,and the brain functional connection of AD patients was more likely to be in a low modularity period and integrated state,and it was easier to transform into a state of whole brain information transmission.In patients with AD,the internal and inter-network connections in the default mode decrease.Pay attention to the increase in the internal and inter-network connections.The decline of AD specialized information processing mode helps to understand the pathological mechanism of AD cognitive changes.