Abstract
Chronic stroke survivors are frequently afflicted with persistent motor impairments. Neural mechanisms underlying recovery and predictors of rehabilitation response remain poorly understood. Advances in MRI-compatible robotic devices have enabled the integration of neuroimaging with targeted therapy, allowing for the real-time assessment of brain plasticity. The present study aimed to identify neuroimaging biomarkers of motor performance and recovery in patients with chronic stroke using functional MRI (fMRI), diffusion tensor imaging and robotic-assisted therapy. In total, 14 patients with chronic stroke (8 women, 6 men; mean age, 55.2±12.2 years) with left middle cerebral artery infarcts from Massachusetts General Hospital (Boston, USA) underwent a 10-week home-based rehabilitation program using an MRI-compatible robotic hand device. Motor outcomes were assessed using the Fugl-Meyer assessment for upper extremity (FMA-UE), Box and Block Test (BBT) and grip strength. Imaging data from 210 sessions were then analyzed to evaluate the degree of task-related brain activation and white matter integrity. Generalized linear mixed models revealed that focused activation in the ipsilesional primary motor cortex (M1) was positively associated with BBT (B=7.57; P<0.001) and grip strength (B=8.65; P<0.001). By contrast, activation in the contralesional ventral premotor cortex was found to be negatively associated with motor outcomes (B=-2.89; P<0.001). Higher fractional anisotropy (FA) in the corticospinal tract and posterior limb of the internal capsule was positively associated with motor performance (FMA-UE, B=133.10; P<0.001), whilst higher FA in the posterior corona radiata was negatively associated with motor performance (BBT: B=-1,409.10; P<0.001). Rehabilitation-induced improvements were also associated with increased ipsilesional M1 activation (B=0.67; P=0.002) and recruitment of the contralesional dorsal premotor cortex (B=3.43; P<0.001). In conclusion, these data suggest that recovery from chronic stroke is supported by lateralized motor network engagement and preserved white matter integrity. Therefore, neuroimaging biomarkers may be exploited for guiding personalized rehabilitation strategies and predicting the patients' response to rehabilitation.