Brain network reorganization and prediction of motor recovery after a stroke

Abstract

Physiological effects related to brain injury such as stroke need to be assessed in all distributed complex networks related to the damaged region. The brain systems have a feature of complex network. Therefore, even though the brain’s structure is damaged by a focal lesion, the damage due to stroke diffuses through the brain network and influences the function of distant brain regions. In this dissertation, in order to gain insights into brain network reorganization during stroke recovery, changes in the motor network were investigated to delineate the role of brain regions on the recovery of motor functions by using longitudinal resting-state functional magnetic resonance imaging data. Differences in plasticity of motor network during recovery were also investigated in different severity groups. Furthermore, prediction of recovery was also investigated. Recovery prediction helps clinicians making individually-tailored rehabilitation plans and allows patients to set realistic goals. In this dissertation, a prediction model capable of measuring the quantity of recovery was proposed using graph theoretical and lesion network analysis. The proposed predictor and motor function recovery were significantly correlated in the supratentorial lesion. Our prediction model showed high performance in the prediction of recovery at three months post-stroke (P=3.67e-16, $R^2$ =0.788). Cross-validation was also performed (P=2.06e-14, $R^2$ =0.746, RMSE=13.15). We expect that the findings and the proposed prediction model can help the planning of patient-specific rehabilitation for stroke patients.