Brain Functional Analysis And Functional Brain Networks Construction For First-episode Major Depressive Disorder Based On Resting-state FMRI

Brain is considered as the most complex system of human body.Human exploration of brain never seem to cease.However,our understanding to it remains limited due to the lack of appropriate measurements and algorithms.As the increase of social and personal pressure,the risk of suffering major depressive disorder(MDD)or other nervous system diseases have been increased recently.The incidence of MDD has become higher and higher.But the pathogenesis of MDD is not fully understood by now,and the diagnostic basis of it is still insufficient.Therefore,it is of practical significance to investigate the brain functional changes of patients with MDD,which will provide great help for MDDs clinical diagnosis.Based on resting-state functional Magnetic Resonance Imaging(rs-fMRI),this thesis investigated the brain functional changes,the non-linear correlation calculation algorithm and the functional brain networks construction for MDD group and health control(HC)group.The specific work is as follow:1.The analysis of brain functional changes for MDD group and HC group.Aiming at the problem of inconsistencies in current research methods and final results,this thesis calculated and analysed the brain functional changes for MDD group and HC group with region homogeneity(ReHo),amplitude of low-frequency(ALFF)and fraction ALFF(fALFF)methods,respectively.Then the brain functional changes for these three methods was compared and discussed.Based on the results,ReHo and fALFF methods showed better performance than ALFF as ALFF was sensitive to the physiological noise.The results of this thesis showed significant brain functional differences in amygdala,lenticular nucleus,temporal lobe,frontal lobe,cerebellum and fusiform gyrus.Also,the functional changes in posterior central gyrus and caudate nucleus for MDD group were first discovered from functional perspective.2.The construction of functional brain networks(FBN).Recent years,researchers has utilized graph theory methods to construct FBNs for MDD or other diseases.However,most of the traditional methods used to calculate the correlation between brain regions were linear algorithm.Based on this condition,a novel nonlinear correlation algorithm known as correntropy was proposed in this thesis to estimate the relativity of FBNs for MDD group and HC group.The FBNs of MDD group and HC group calculated by three linear correlation methods,which include pearson,spearman and kendall,and correntropy method were measured.Compared to linear algorithms,the small-world values of both MDD group and HC group for correntropy algorithm were much higher.The results showed the superiority and effectiveness of correntropy algorithms we proposed.In this thesis,the functional brain changes were analysed and FBNs were constructed for MDD group and HC group.Based on the experiment results,this thesis provided a new objective way to diagnose MDD in clinical.Also,a novel nonlinear correlation calculation algorithm called correntropy was introduced to provide a new approach to construct FBNs instead of the traditional linear algorithms.The results showed that the method we proposed performed better than all other linear methods.