Probing Neural Compensation in Rehabilitation of Acute Ischemic Stroke with Lesion Network Similarity Using Resting State Functional MRI

<b>Background/Objectives</b>: Neural compensation, in which healthy brain regions take over functions lost due to lesions, is a potential biomarker for functional recovery after stroke. However, previous neuroimaging studies often speculated on neural compensation simply based on greater measures in patients (compared to healthy controls) without demonstrating a more direct link between these measures and the functional recovery. Because taking over the function of a lesion region means taking on a similar role as that lesion region in its functional network, the present study attempted to explore neural compensation based on the similarity of functional connectivity (FC) patterns between a healthy regions and lesion regions. <b>Methods</b>: Seventeen stroke patients (13M4F, 63.2 ± 9.1 y.o.) underwent three resting-state functional MRI (rs-fMRI) sessions during rehabilitation. FC patterns of their lesion regions were derived by lesion network analysis; and these patterns were correlated with healthy FC patterns derived from each brain voxel of 51 healthy subjects (32M19F, 61.0 ± 14.3 y.o.) for the assessment of pattern similarity. <b>Results</b>: We identified five healthy regions showing decreasing FC similarity (29-54%, all corrected <i>p</i> < 0.05, effect size η²: 0.10-0.20) to the lesion network over time. These decreasing similarities were associated with increasing behavioral scores on activities of daily living (ADL, <i>p</i> < 0.001, η² = 0.90), suggesting greater neural compensation at early-stage post-stroke and reduced compensation toward the end of effective rehabilitation. <b>Conclusions</b>: Besides direct FC measures, the present results propose an alternative biomarker of neural compensation in functional recovery from stroke. For sensorimotor recoveries like ADL, this biomarker could be more sensitive than direct measures of lesion connectivity in the motor network.