Dynamic Resting-State Functional Connectivity Patterns Of Depression: A MEG Study

Interactions between brain regions and networks of resting-state are proved to be unstable and variable. Dynamic characteristics of resting-state functional connectivity could be used to explore pathogenic mechanism of depression. Recently, deviant subtle network dynamics of functional connectivity have been related to dysfunctional cognitive process. Abnormal instantaneous "Quasi-stable’ functional states of large-scale neural activity and their variation during resting-state have been observed in Alzheimer’s disease and schizophrenia. However, characteristics of these momentary dynamic patterns of depression still remained unknown. Hence, we seek to find characteristics of these transient functional states of depression with Magnetoencephalogram (MEG), which has a high temporal resolution.Firstly, this paper focus on spontaneous functional states of several brain areas related to depression. We choose time series of alphal band, during which the RSN were the most active. Wavelet coherence is used to calculated dynamic functional connectivity of brain regions of interest (ROI). Then, an improved Spherical K-means algorithm (SP-Kmeans) is applied to classify the functional connectivity states over the time period. As a result, one extra state of the patient group is discovered. Configuration of functional connectivity, duration, occurrence and percentage of the rest matched states all suggest different features of the depression compared with the controls. One extra state of the depression demonstrates more frequent occurrences and shorter duration. Dynamic functional connectivity of the patients exhibits more unstable and complex. Moreover, maps of patients’ functional states all demonstrate obviously modularity of brain areas within the Default-mode network (DMN), Frontal-parietal network (FPN) and the Salience network (SN). Abnormal characteristics of states embody disorganized information interaction of the depression.After that, we investigate further the features of these three highly modularized networks. Synchronization likelihood algorithm is applied to calculate long-term coupling within the three networks separately. SP-Kmeans clustering is also used to explore functional states and changing rule of DMN, FPN and SN separately. Results show that DMN play a leading role in dynamic changes of states, remain relatively stable and influence rapidly changed FPN and SN in a relatively long period. We hypothesis that DMN act as a core network, providing stable maintenance of the whole-brain microstates. FPN has several functional states shifting frequently, which suggests a flexible response of FPN to short-term neural activity. Features of these networks and their periodic variability regulation in a long time infer special modulation effect of them.Whole brain microstate is consisted of multiple sub-networks and modulated by interaction and self-variation of them.In conclusion, we explore deviant functional states dynamic characteristics of depression during resting-state from multiple perspectives. These transient "quasi-stable" functional states are supposed to be closely related with dysfunction of the depression in spontaneous brain processing. They may help us to deepen the understanding of the neurophysiological mechanisms of depression.