The Hurst Exponents Of The FMRI Resting-State Networks And Dynamic Interactions Among The Networks In Depression

Previous studies have reported the abnormal activities of the depression-related brain regions, disturbed between-regions connectivity and the dysfunctional networks, which could help our understanding of the pathophysiological mechanism of the depression. However, all these were based on the static performance of functional magnetic resonance imaging signals. In the resting-state, the interactions of resting-state networks would exhibit dynamic changes accompanied with the task-free brain activities. Exploring such dynamic changes could mirror the intrinsic brain activities in detail. The paper aimed to investigate the dynamic abnormality of the brain intrinsic activity via the associated analysis to the dynamic anomaly of the individual network and inter-network dynamic interaction. Further study was conducted with the exploration of the dynamic communities of the resting-state networks and the change of the functional state patterns by introducing the dynamic community detection method. The dynamic research provided a new understanding of the neuropathological mechanism of the depression, and important information for the clinical diagnosis of the depression. The detailed work involved:(1) The Hurst exponents of the resting-state networks in depressionThe spontaneous activity of the brain has been reported to be scale-free, and probably related to the cognitive and mental disease. Hurst exponent was used to describe such scale-free property, and reflected the dynamic activity to some extent. Taking the Hurst values of the resting-state networks as features, the support vector machine was used to identify the depressed patients from the healthy controls. The depression patients could be distinguished with sensitivity as high as 95%. The abnormal resting-state networks were explored by the weight vectors within the classifier. As compared with the healthy controls, the depressed patients exhibited decreased Hurst exponents in the right fronto-parietal network and default mode network, implying that the patients might efficiently process information with the negative bias. The salience network, left fronto-parietal network and the ventral prefrontal network with increased Hurst values, might work as compensatory mechanisms for dispending from the negative bias.(2) The abnormal dynamic interactions among the resting-state networks in depressionIn spite of the anomaly of the functional connectivity within the resting-state networks, the depressed patients show the abnormal connectivity among these networks as well. Thus, it is necessary to explore the dynamic interactions between networks and the associations with the dysfunction of the individual network. The results of this study demonstrated the abnormal dynamic interaction involving the default mode networks in the depression. The dysfunction of default mode network might be associated with the dynamic interactions, via which the default mode network was frequently subjected to the dynamic modulation from the salience network and ventral prefrontal network, while exerted frequent influence on the fronto-parietal network.(3) The analysis of the dynamic community in depressionThe dynamic community detection method was applied to explore the dynamic modules, networks’ flexibility and the extent of their participations in multiple communities. To our knowledge, it was the first study to explore dynamic community of resting-state network in depression. We calculated the inter-slice couplings between the adjacent time windows, and kept the information of the dynamic brain activity change at the switching point between the windows. Results suggested that the flexibility and promiscuity of the salience network was decreased in depression, implying decreased functional module participation and frequency of the changes among the modules in the depressed patients. The salience network might be in a dominant place in the mechanism of the functional networks in depression.(4) The change of the functional state in depressionResting-state activity exhibits a series of instantaneous mental states and the switching between them, accompanying with the process of the micro-information. The dynamic change of the functional modules of the each resting-state network acquired by the dynamic community detection method could mirror such changes of the instantaneous mental states. The mutual information was used to describe the change of the functional modules at adjacent time windows, and the minimum value of the mutual information was used to estimate the time point when the largest change of the modules happened, which was defined as the change point of the functional state. The jump matrix was introduced innovatively to conduct a tentative study for the mechanism of the change of the functional communities under the change point. The results showed that the depressed patients possessed the change of the functional state similar to that of the healthy controls. However, the reasons for urging the change of the functional state were different between two groups. In depression, the interactions between the functional communities, which dominated the change of the functional states, were located in the communities where the posterior cingulated gyrus and inferior frontal gyrus, the insula and precuneus were assigned. Meanwhile, results demonstrated that some resting-state networks might spend more time dwelling in certain functional communities, implying the relation to the fact that the depressed patients were excessively immersed in negative bias.