The Research Of Depression's Functional Brain Network Features Classification Based On FMRI Data

There is a relatively clear diagnostic criteria and treatment of major depressive disorder in modern medicine,which are commonly used method is based on the diagnostics for diagnosis and treatment.However,this approach is still in the primary research phase,especially there is still no a reliable conclusion of etiology and pathogenesis research about major depressive disorder.Now mostly diagnosis is according to the patient's family members to describe the major depressive disorder's behavior at ordinary times,and coupled with the doctor's own observations and the corresponding medical knowledge to assess.This approach puts forward a very high requirements to the doctor's diagnosis ability and work level,at the same time as the main sex is too strong also leads to misdiagnosis and delayed the best time of treatment,so that lead to further deterioration.With the development of modern technology,especially the brain Imaging technology,it provides a very convenient method to diagnosis and treatment of brain disease,in particular the functional Magnetic Resonance Imaging(fMRI)technology have played an important role in promoting major depression treatment.Functional magnetic resonance imaging(fMRI)has its own unique advantages,as a means of research of brain dysfunction,it can be a specific research,which can under the condition of no wound to the brain.And the fMRI technology is widely spread throughout the world as well as to promote the medical research on major depression treatment.As Biswal combined the fMRI technology with resting state in brain function research,the resting state fMRI research began to come into the mind of the people,especially in the study of brain disease.Depression is always been the hot spot of the research on medicine,at the same time because it bring heavy burden and serious impact tohis patients,patients' families and society,so it is becoming more and more attention from society.But the diagnosistechnique that for depression and other mental illness is relatively single.We need to explore an efficient and systematic method.Among various kinds of brain disease,using the resting state fMRI technology research have been achieved good results and has played a huge role in clinical diagnosis.The resting state functional magnetic resonance imaging data has provided a new perspective and method for the research of mental illness.Considering the current research situation,we use the depression group and healthy controls to relevant functional magnetic resonance imaging data to assay.we select some significant differences among the attributes to construct the corresponding classifier model.By choosing the appropriate classification algorithm to build the appropriate classification model,which can effectively to discriminate the depression group and health control group.And it can be a important significance for building the physiological indicators of clinical and also for building auxiliary diagnosis of depression model.We combine complex network theory system with brain science theory system for interdisciplinary research,and we use functional magnetic resonance imaging technology to get the patients' brain images,and by using fMRI data to construct the functional brain network,and use complex network theory to study the abnormal brain network topology structure.We select the feature attribute and use the classification algorithm to classify the patients and the control.In this perspective,we want to provide a more comprehensive,accurate research direction for the sick of depression.In this paper,the main work is as follows:1.Using functional magnetic resonance imaging technique to acquire the resting state fMRI image.Task in subjects under the condition of resting state for fMRI data.In our experimental study,the subjects included patients with major depressive disorder(36),and normal person(37).2.To preprocessing the data.We use the toolbox of MATLAB to preprocessing the data,such as DPARSF and REST.The step has time calibration,dynamic calibration,space,standardization,smooth,low frequency filter.3.To build the brain network and calculate the network features.The brain is segmented into 90 brain regions by AAL template,we use the nonparametric permutation to test the attributes of the network contrast of the 90 brain regions,then choose a difference larger brain areas for research.And we also calculate the network attribute characteristic features,including a network costs,global efficiency and local efficiency,efficiency of node and edge betweenness.Those will be as the classificationcharacteristics to research.4.Study on the machine learning methods,and the classification algorithm that suitable for our experiment.5.To classify the fMRI data.We use the classification algorithm and the network characteristic to construct classification model and to forecast experiment,eat last analyze the result of the experiment.