A Study On Dynamic Characteristics Of Functional Brain Networks Based On Resting-state Functional Magnetic Resonance Imaging In Major Depressive Disorder

As a prevalent psychiatric disease,major depression disorder(MDD)is characterized by emotional and cognitive dysfunction,which has a significant impact on patients’ life and has been reported to be a severe economic burden to the society.Previous research leveraging functional magnetic resonance imaging(fMRI)technology found that MDD and other mental diseases are related to abnormal brain activity.MDD has been considered as a brain network disorder.A large number of neuroimaging studies on MDD have shown that patients with MDD may be associated with aberrant activity and connectivity of specific brain networks.For example,dysfunction of brain connectivity in the default mode network(DMN)may cause excessive thinking and rumination in MDD patients;enduring negative emotions in MDD patients were associated with abnormalities in the affective network(AN)of the brain;the reward network(RN)was found to be related to the core symptom of anhedonia in MDD patients.Abnormality in the cognitive control network(CCN)may lead to failure in regulating the emotions.At present,the majority of traditional fMRI studies on brain networks used static functional connectivity analysis.Static functional connectivity(SFC)analysis treats the brain as a stationary system,thus actually calculates the mean strength of functional connectivity.However,the method of SFC has some limitations.In recent years,some studies have provided more evidence that the functional connectivity of the human brain is highly dynamic which represents the flexibility of functional coordination between different brain regions.The dynamic characteristics of brain connectivity have been accessed in terms of analyses on spatial patterns as well as temporal patterns.By analyzing co-activation patterns(CAPs),researchers have investigated the spatial patterns of dynamic brain connectivity.In addition,dynamic functional connectivity(DFC)analysis methods like the sliding window algorithm have accessed the temporal characteristics of dynamic functional connectivity.Although traditional SFC studies have shown that MDD may be associated with specific brain networks,a lot of problems were left to be solved.First,it is not clear at present about the differences in the spatial patterns of dynamic brain connectivity of specific brain networks between MDD patients and healthy subjects.Secondly,the temporal patterns of dynamic functional connectivity in MDD subjects and healthy subjects remain unclear.In addition,quantitative identification of MDD patients using the changes in the temporal and spatial patterns of the brain dynamic characteristics of MDD patients was the next step to explore whether machine learning can help clinical diagnosis of MDD patients.Therefore,the present study uses fMRI-based dynamic brain connectivity analysis to discover the differences in spatiotemporal patterns of brain activity between MDD patients and healthy subjects,and to quantify the temporal and spatial patterns of abnormal brain dynamics in MDD patients,which may provide deep understanding of neurobiological mechanism in MDD.The main work is as follow:The first chapter used CAPs analysis to explore the differences in the complexity of the internal fluctuations in brain activity between MDD and healthy subjects.A functional connectivity analysis based on independent component analysis(ICA)was used to extract the time series of the MDD patients and the healthy subjects of the DMN,left and right CCN and SN brain networks,then sorted by the amplitude of blood oxygen level dependent signal,after which the analysis of CAPs based on K-means clustering was conducted.The results showed that the changes in spatial CAPs over time in the three brain networks were different in MDD patients compared with healthy subjects,which suggests differences in spatial patterns of brain dynamics in specific brain networks between MDD patients and healthy subjects.The second chapter is to explore the differences in temporal expression patterns of brain dynamics between MDD patients and healthy subjects.The FLS-based dynamic functional connectivity analysis was combined to explore whether the temporal expression patterns of the dynamic characteristics of the brains of MDD patients and healthy subjects are different,and further to explore whether the gender differences have an effect on the temporal expression patterns of the brain dynamic characteristics of MDD patients.The results showed that the temporal patterns of brain dynamics in the amygdala of MDD patients and healthy subjects were different.The spatial pattern of brain dynamics in the amygdala of MDD patients is greater than that of healthy subjects,and the temporal patterns of brain dynamics in the amygdala of MDD patients of different genders showed different states.The results indicate that the neuronal activity in the amygdala of MDD patients and healthy subjects may show different states over time.In the third chapter,we combined the analysis of brain dynamic functional connectivity with machine learning.The low efficiency of clinical diagnosis of MDD has been hindered the development of clinical diagnosis and treatment of MDD patients.With the development of machine learning,people began to explore this problem.It was found that abnormal functional connectivity can be used as an imaging feature,providing new opportunities for the accurate diagnosis of MDD.Therefore,we hope to explore the influence of brain dynamic characteristics and static characteristics on the classification efficiency,verify the advantages of brain dynamic characteristics,and try to improve the efficiency of machine learning classifier according to the changes of temporal and spatial patterns of MDD patients’ brain dynamic characteristics.The results show that compared with static functional connectivity analysis,the classification model based on dynamic functional connectivity is better.In this thesis,according to the difference of dynamic brain function connectivity between MDD patients and healthy subjects,the difference of dynamic functional connectivity in temporal and spatial distribution was researched,and quantitative identification of major depression based on resting-state dynamic functional connectivity was conducted by machine learning.This article provide a framework to forming a complete analysis system of abnormal brain dynamic network between MDD patients and healthy subjects.