Brain Network Analysis Based On Bayesian Approach

The brain of human beings is one of the most complicate system in nature. The differentiation and integration of function are the principles of human beings brains’ function, whose execution always relies on the extensive effects among many brain areas. Before lots of the research of neuroimaging offer forceful arguments,which proves that brain networks of schizophrenia patients have many informal connections. However, Whether these informal connections will lead to topology of brain networks, so it is necessary to research the function of brain from the perspective of the network as a whole. And graph theory is an important tool of describing the characters of network. For the structure of the human brain networks, many research has proved that it has the character of small world."Small world" is a vivid definition of network topology as follow:In the network, most of the nodes is not connect with each other.,but most of the nodes can get into through any other nodes with a few steps. The analysis of the complex brain network based on graph theory is becoming an issue of the research of neurosciences at present.This article is aimed at using function Magnetic Resonance Imaging technology to research the informal of the brain functional networks at the pathology of schizophrenia, so that we can deepen the understanding of schizophrenia. In this paper, the main work focus on several aspects:The brain of human beings can be divided into90brain regions according to the AAL template, estimating the activated situation of Each participant90brain regions using the generalized linear model. And using Bayesian estimation that gets the the brain’s joint activation probability of connection between the two regions. It estimates the connection of two regions by constructing statistic in order to construct the connection network of brain function which adjust to Schizophrenic patients and healthy people. It also inspects these two network based on the nature of the graph theory which focus on two aspects:Firstly, in view of the node properties. Secondly, In view of the network overall organizational attributes. And it concentrates on the measure of small world in the network. In this paper, there are two aspects of results:Firstly, For healthy people, the patient group of the network node properties change a lot. Through calculating node properties,we find that the degree of most regions, the whole effectiveness and the part effectiveness have big differences. Secondly, schizophrenia patients and normal people remain the characters of small world to the brain network. And the nature of patients’small world is not diminished but slightly increases for normal people.