Dynamic Network Interaction For Identifying Bipolar From Unipolar Depression Via Resting-state FMRI

Misdiagnosis of bipolar depression(BD)as unipolar depression(UD)may cause very deleterious consequences for treatment,especially in the early stages of the disease.At present,numerous clinical measures have been taken to identify the risk of transforming into BD.However,they still lack sufficient specificity and rely heavily on the subjective judgment of doctors.Although the results were inconsistent,the development of neuroimaging made it possible to find biomarkers for the identification of BD and UD.However,previous studies on the identification of BD and UD were mostly based on the stationary hypothesis of signals,and the ability of the dynamic network interaction in the resting state in the identification of the two diseases has not been studied.The main work of this paper is to explore the difference in brain functional network interaction between patients with BD and UD using the functional magnetic resonance imaging(fMRI)in resting state.The main research contents are as follows:1.First,we studied the difference between resting state fMRI dynamic network interaction in patients with BD and UD.Dividing the whole brain into 6 functional networks,the dynamic interactions between networks were obtained based on the sliding time window method.The differences in network dynamic interaction between patients with BD and UD were compared using permutation test.The results showed that the dynamic network interactions of cognitive control network(CCN)significantly decreased in BD.In the further classification,the dynamic network interaction has the better performance than the static characteristics.This suggests that the dynamic interaction of the resting state network can better captures the difference in brain network damage between BD and UD patients.2.Based on the above results,further research was carried out to characterize alterations in specific brain networks for depressed patients who transformed into BD(t BD)from UD.The module allegiance(MA)from resting-fMRI by applying a multilayer modular method was estimated in patients with tBD,BD,UD and healthy controls(HC).A classification model was trained on t BD and UD patients.HC was used to explore the functional declination patterns of BD,t BD and UD.The result shows that: classification performance was acceptable and the difference mainly reflected in the abnormality of internal interaction in the default-mode network(DMN).Compared with HC group,we found that both the BD and t BD groups focused on the injury of somatomotor network(SMN),while UD group was mainly reflected in the dysfunction of DMN.Notably,the abnormal patterns of brain network between patients with BD and t BD well overlapped,except for CCN.In summary,these findings suggest that despite the similar clinical symptoms,BD patients or t BD patients and UD patients had differences in the spatiotemporal patterns of the whole brain network.These may serve as a biomarker for the identification of BD patients from UD patients in their early depressive stage of disease.