“…We also performed data format conversion, head motion eddy correction, gradient orientation correction, brain mask acquisition, and tensor calculation using the Functional MRI of the Brain Software Library version 6.0.3 commands to generate FA, MD, AD, RD, S0, and other files. 9,16,17 CST tracking and extraction were performed using the MATLAB platform and open-source AFQ software (https:// github.com/yeatmanlab/AFQ). The process mainly included (1) converting the data obtained from the previous steps into a dt6 file in MATLAB format using the dtiMakeDt-6FromFsl command; (2) setting the tracking parameters for FA threshold greater than 0.2 and turning angle less than 45° for whole-brain fiber tracking; (3) segmenting 20 major fiber tracts from whole-brain fibers; (4) cleaning up the raw fiber tracts based on the average length or 4 standard deviations of fiber bundle cores to clean up the original fiber bundles; (5) equally dividing the CST into 100 subsegments from the midbrain to the primary motor cortex and extracting diffusion tensor values for each segment; to comprehensively assess the microstructural changes in the CST during stroke rehabilitation, we computed the FA, MD, AD, and RD values for each subsegment, as these indexes provide distinct perspectives on fiber bundle damage 18 ; (6) to facilitate inter-group comparisons, the data of the subjects with right hemisphere cerebral infarction were flipped left-right; consequently, the left-side data of all patients represented the ipsilesional side, while the right-side data represented the contralesional side; correspondingly, accounting for the potential systematic effects from hemispheric differences, for the HC group, a new set of data, including bilateral corticospinal data, was generated after flipping the left and right hemispheres; and (7) extracting CST information, with the left CST being ipsilateral to the lesion and the right CST being contralateral to the lesion (Figure 1).…”