Longitudinal Changes Of Brain Anatomical Network In Depressions

Major depressive disorder (MDD) is a common mental disorder. Although various endeavors from neurobiochemistry, heredity and social mentality have been made to investigate the pathogenesis and etiology of MDD, its pathogenesis is still not completely clear. Benefited from the developing neuroimaging technology and closely combined multi-disciplines, MDD has been recognized as a disconnection problem. Herein, we aim to investigate the brain topological infrastructure changes on MDD patients before and after antidepressant treatment from the aspect of brain white matter networks based on the diffusion tensor imaging (DTI). Meanwhile, the correlations between these changes and the clinical behaviors have also been explored. In total, our work includes the following four parts.(1) From the segmentation, integration and centrality aspects of the brain structural network, we analyzed the network topological properties of depressionsWe started at constructing the F A-weighted undirected white matter network, and then calculated the topological indexes. As a result, no significant changes had been observed in the global properties, while the damaged nodes in the local properties were mainly in both cognitive-emotional circuitry and frontoparietal circuitry. Moreover, the centralities in the right prefrontal cortex involving the right middle frontal gyrus and the right gyrus rectus were negatively correlated with the duration of disease. Additionally, the centrality in the right inferior frontal gyrus (triangular part) and the efficiency in the right superior frontal gyrus (orbital part) were both positively related to the depression severity. These findings first suggested that altered connectivity involved in affective and cognitive processing procedures might contribute to the pathogenesis of MDD.(2) Inspired by the economical efficiency property of healthy brain structural networks, we examined the impact of the high cost hub nodes on recognizing patients with MDDWe took the network indexes as the classification features and divided these features as HUB and MIX feature pools. MIX indicated all the features, while HUB meant the features only marked as hubs from M IX’s feature pool.Wet hen e mployed the S VM c lassifier w ith RBF ke rnel to classify MDD patients from the healthy controls. The results showed both HUB and MIX features were effective. And the classification performance of HUB was better than that of MIX, which indicated that the damaged hub nodes had larger influence on the entire network. In the further discriminative feature analysis, we found that most of the HUB consensus features were in the frontalparieto circuit, suggesting that the hubs could be served as a type of valuable potential diagnostic measure for MDD, and the hub-concentrated lesion distribution of MDD was associated with the pathology mechanism of MDD.(3) Differences in the patterns of brain structural networks of depressions during episodes and remission periodWe proposed a novel classification framework in order to better describe the pathological information of MDD. Specifically, we introduced five physiological parameters and computed five kinds of weighted white matter networks in order to better describe the pathological information of MDD. Three representative network indexes including the node strength, communicability a nd edge betweenness centrality were selected as the classification features respectively. The results exhibited the similar deficit patterns in both current MDD (cMDD) and remitted MDD (rMDD) relative to healthy controls, which involved the salience network (SN), default mode network (DMN) and frontoparietal network (FPN). Meanwhile, the different patterns between cMDD and rMDD were intra-communicability within DMN and inter-communicability between DMN and the other sub-networks including the visual recognition network (VRN) and SN. These findings implied that the impairment of MDD was closely a ssociated with the alterations of connections within SN, DMN and FPN, whereas the remission of MDD was benefitted from the network compensatory of intra-communication within DMN and inter-communication between DMN and the other sub-networks (i.e., VRN and SN).(4) A follow-up study of hub-level community structure network (HLCSN) changes of MDD patients before and after antidepressant treatmentThe segregation and integration of neural information were very important aspects of the information communication among the whole brain, and the balance between them was essential for the manipulation of distributed networks underlying cognitive function. The hubs and communities could measure these two aspects respectively. Few study has been investigated the effect of antidepressants on the segregation and integration of information communication in brain network. Therefore, w e explored the network pattern changes of depressions before and after antidepressant treatment from the segregation and integration point of views by using DTI. Regarding to the method, we modified the permutation network framework to adapt our data structure which could statistical analyze the between-group difference at the individual level. We began with constructing the brain white matter networks and performing the module detection. Then we focused on the analysis of HLCSN for each subject. It was detected that the MDD patients before and after antidepressant treatment was quantitatively reconfigured and most of the alterations anchored within the frontoparietal and cognitive-emotional circuit. Moreover, the shift of hub roles between the healthy and MDD patients both before and after antidepressant treatment were changed from provincial hubs to connector hubs or connector non-hubs within the frontoparietal circuit. The interconvertible shift of hub role between pre-and post-treatments was observed in the left INS. Additionally, the correlative results exhibited that the modular measures in the left ACG and right PCUN was correlated with the 17-item HAMD score. Meanwhile, the reduction in inter-module degree of the right SPGmed was positively correlated with the reduction in the 17-item HAMD score after antidepressant treatment. These findings presented that antidepressant treatment drove the reconfiguration of HLCSN within fronto-limbic circuit, which suggested a disbalance between the segregation and integration of the brain system and the dysfunction of integration of the different brain functions in MDD.These abnormalities might be a potential predictor of antidepressant curative effect, especially of SSRIs.