Classification Research Of Brain Funcitional Network In AD And MCI

Human brain is a complex dynamic system, is the physiological basis of brain information processing and cognitive expression. Explore the structure and function of the brain, helps to understand the working mechanism of the brain, has great significance to the study of brain-related pathologies.In recent years, many researchers have applied the complex networks theory in brain cognition research. Each brain region regards as a node in a network, connection between two nodes are seen as a edge, brain can be abstracted into a complex brain network. By analyzing the properties of the network, we can understand the workings of the brain network, having geret significance for researching on neurological diseases.With the accelerated pace of population aging, Alzheimer’s disease is becoming an important issue to the health of the elderly. Alzheimer’s disease is a neurodegenerative disease with high incidence rate, there is no effective treatment currently. Through the intervention of its early stage-mild cognitive impairment, we can delay disease progression.Studies have found that brain functional network of Alzheimer’s disease is abnormal, however, changes with the brain network topology attributes in the early stages of the disease has not been explored, the impact on the brain function network with different brain regions divided remains unclear.In this paper, we researched the complex network of Alzheimer’s disease, late mild cognitive impairment, early mild cognitive impairment and normal controls. The main works are shown as follows:(1) Collect resting state fMRI data and structure MRI data of Alzheimer’s disease, late mild cognitive impairment, early mild cognitive impairment and normal controls. Preprocessing of these raw data are necessary to reducing noise and improving the quality of data.(2) Using AAL brain template and Craddock brain tempalte to define nodes, extracting time series of each brain regions, constructiong brain function network of all the subjects, node properties were calculated including degree of node, betweeness centrality, eddiciency of node, then area under the curve of these property. K-S statistical tests carried on the node properties between different groups as classification features, then svm algorithm was used to classifie different two groups.(3) After the structural MRI images were preprocessed, modulated gray images were obtained, it can reflect changes in the volume of gray matter. Analyzed differences between different groups using voxel morphology method. The area of interest with differences between the four groups were used as classification features; mini-mental state examination of behavioral scale was selected as classification features.(4) SVM method was used to construced a classification model with the area under the curve and the gray volume feature and mini-mental state examination.Classification results showed that only the brain function network characteristics in two brain divided way, the divided way was more fine, the accuracy was more higher. After joining gray characteristic, the accuracy can be significantly improved; after joining scale characteristics, the accuracy of classification was not highly improved; three kinds of features are classified together, classification effect is the best.This means that the brain functional network method combining gray and scale can be used to study the brain disease, provideing an auxiliary means for clinical diagnosis.