2021

Liu, Wenya

Major depression disorder (MDD) is a prevalent psychiatric disorder globally, affecting one in six people. From the view of theoretical models, the dysconnectivity of functional networks is considered a critical cause in the cognitive and emotional dysfunctions of MDD. However, the pathophysiology of MDD remains unclear due to the non-replicability in terms of methodologies and datasets. Both of the causes of MDD and the human connectome are incredibly complex, and novel experimental paradigms and advanced methodologies are needed to explore the pathophysiological mechanisms of MDD. In this thesis, we explored the altered oscillatory functional connectivity in MDD during music listening conditions and resting states. In the first study, we investigated the frequency-specific static functional connectivity (FC) in MDD during music listening at the sensor level. We found altered FC networks and the non-lateralized effect in the delta and beta bands, and we got the best classification performance in the beta band by the support vector machine classifier. In the second study, we proposed a comprehensive framework to identify the dysconnectivity of oscillatory networks in MDD during resting states at the cortical source level. Fully considering the incomplete consistency in the adjacency and spectral modes between the healthy group and the MDD group and the multiway structure of the constructed data, we first introduced the coupled tensor decomposition (CTD) model for EEG signals recorded during music listening. We identified three hyper-connectivity networks and three hypoconnectivity networks characterizing the dysconnectivity networks in MDD under music perception. Based on the CTD model, we also explored the hyper-and hypo-connectivity networks in MDD during resting states. In the third study, we examined the dysfunction of sensor-level networks in the alpha band. In the fourth study, we explored the source-level dysconnectivity networks characterized by spatio-temporal-spectral modes of covariation in MDD. In conclusion, this thesis investigated potential biomarkers of oscillatory networks and provided promising references to reveal the pathoconnectomics in MDD. The proposed analysis pipeline based on the CTD model can be extended to other psychiatric disorders.