Research On Brain Functional Networks Of Patients With Neuropsychiatric Diseases Based On Graph Signal Processing

Neuropsychiatric diseases are disorders of brain functional networks that can be measured by changes in network metrics.However,for the same functional network metrics,different literatures on the brain functional networks of patients with the same neuropsychiatric disease have reached different conclusions,and in some cases,the conclusions are even opposite.In addition,the existing literatures only analyze the topology of brain functional networks in neuropsychiatric,and do not consider about the signals in brain regions.Based on emerging graph signal processing(GSP),this paper studies the brain functional networks of patients with various neuropsychiatric diseases.The main contributions of this paper are as follows:(1)This paper finds that the selection of connection density will affect the group differences between patients with neuropsychiatric diseases and healthy subjects,and there is a reversal phenomenon in some metrics,which explains the reason why different literatures on the brain functional network of patients with the same mental diseases have reached different or even opposite conclusions for the same functional network metrics.In addition,this paper gives the optimal connection density range for the study of brain function networks of patients with various neuropsychiatric diseases.(2)Analysis of the intergroup differences of patients with various neuropsychiatric diseases and healthy subjects is performed based on eight complex network metrics,and the average results for each network metric of each group of samples within the range of optimal connection density are given.(3)Multi-scale analysis of brain functional networks in patients with neuropsychiatric diseases is performed using SGWT.The spectral wavelet transform filters are designed,and the corresponding spectral wavelet coefficients of each brain region at each scale are calculated according to the graph Laplace matrix of each subject and the blood oxygen level dependent contrast signal on each brain region of each subject.By analyzing the between-group differences at different scales,this paper finds that there are abnormalities in the energy distribution of brain functional networks in schizophrenic patients under different frequency bands,and gives the specific abnormal brain regions under each frequency band.