Structural and EEG motor networks distinguish level of motor impairment after stroke

The primary objective of this research is to elucidate brain connectivity patterns that are associated with motor impairment in individuals recovering from stroke. Here we estimate individual patient (1) structural connectivity using stroke lesions modeled from MRI and (2) brain functional connectivity from EEG using graphical modeling informed by the structural connectivity of individual patients. Our analyses concentrated on the brainstem and three motor-related cortical areas in both hemispheres: precentral cortex (primary motor cortex), caudal middle frontal cortex (dorsal premotor cortex), and superior frontal cortex (incorporating the supplementary motor cortex). Network analysis metrics were applied to structural and functional data, including degree centrality, average efficiency, and betweenness centrality. These measures were used to contrast motor networks in patients divided into two groups based on their upper extremity motor deficits. We found that structural disconnection of the brainstem from both dorsal premotor cortex and primary motor cortex indicated the highest probability of low motor status. All structural connectivity metrics of the 3 ipsilesional motor areas were higher in the higher motor status group. We integrated structural connectome with source localized EEG data to estimate the network analysis metrics. The functional connectivity revealed a notable pattern similar to the structural connectivity in the band (1-3 Hz) within ipsilesional motor areas. In addition, contralesional motor areas showed enhanced connectivity in the band (14-29 Hz) in patients with higher motor status, possibly as a result of compensatory neural plasticity of the contralesional homologues. We also introduced a novel metric, motor betweenness centrality, to quantify the role of non-motor regions in rerouting motor-related communication for both structural and functional network analyses. Almost all new hubs for routing the communication with motor-related areas identified in the structural connectome were associated with lower motor status, suggesting network reorganization in patients with greater impairment is inefficient. However, EEG-based routing analyses further revealed that high motor status patients engaged nearby cortical areas, while low motor status patients rerouted signals through more distant and less efficient regions. EEG connectivity metrics can potentially be measured at multiple points during treatment and may reflect functional plasticity. In summary, the structural and EEG-based functional network properties degree centrality, efficiency, and betweenness centrality reliably identified key motor ROIs predictive of motor status. Motor betweenness centrality captured plasticity-related rerouting outside primary motor regions. Delta and beta band EEG networks were most informative of motor recovery. There was notable overlap between the structural and functional connectivity findings. In particular, ipsilesional motor areas, especially M1 and PMd, were consistently identified as critical hubs in both modalities. Structural disconnection of these areas predicted poor motor status, while higher EEG connectivity in these regions in and bands predicted better outcomes. These findings underscore the clinical potential of combining multimodal network measures to monitor, predict, and guide recovery in stroke rehabilitation.