Abstract:Effective description of the brain function after stroke is the key to accurate rehabilitation assessment, and it is of great significance to explore the nonlinear complexity characteristics of the brain from the perspective of complex networks. In this study, we investigated the brain functional connectivity alterations after stroke by constructing a multilayer network model. Firstly, we obtained multichannel EEG signals in different frequency bands (θ, α, β and γ) during the multi-joint compound movement. Furthermore, we introduced the weighted phase lag index (wPLI) and KL-modulation index (MI) to construct the within-frequency subnetworks (WFN) and cross-frequency subnetworks (CFN), respectively. Then multilayer network was constructed by the aforementioned subnetworks. Calculating multilayer participation coefficient (MPC) and multiplex clustering coefficient (MCC) to explore differences in connection strength within subnetworks. The algebraic connectivity was used to compare the differences in multilayer network topology from global perspective. β frequency band WFN showed significantly stronger connectivity in healthy group compared with stroke patients. Conversely, the θ-γ CFN in patients exhibited significantly higher connectivity strength compared with controls, while the opposite was true for α-β CFN. There were significant differences in network nodes between the left and right brain regions in controls, whereas the distribution of MPC in both hemispheres was evenly distributed in the patients. Global metrics indicated that the algebraic connectivity of the patients’ brain network was significantly lower than that of the controls. These findings have important implications for understanding the brain functional connectivity in stroke and developing effective rehabilitation and therapeutic strategies.

Brain Functional Alterations of Multilayer Network after Stroke: A case-control study based on EEG signals