Sensorimotor Network Connectivity in Stroke

A region of interest (ROI)-to-ROI approach was adopted to investigate the resting-state sensorimotor network primarily. Twelve ROIs (with corresponding MNI coordinates) with 6-mm radii were predefined for the paretic hand representation from a meta-analysis of movement-related fMRI in 472 patients with various impairment from acute to chronic phase after ischemic stroke [13 (link)], including contralesional M1 (cM1, − 38, − 24, 58), ipsilesional M1 (iM1, 42, − 14, 52), contralesional S1 (cS1; − 36, − 30, 60), ipsilesional S1 (iS1; 40, − 28, 52), contralesional supplementary motor area (cSMA; − 4, − 6, 54), ipsilesional SMA (iSMA; 4, − 6, 54), contralesional dorsolateral premotor cortex (cPMd; − 42, − 10, 58), ipsilesional PMd (iPMd; 42, − 6, 56), contralesional ventrolateral premotor cortex (cPMv; − 46, − 10, 48), ipsilesional PMv (iPMv; 42, − 6, 48), contralesional anterior intraparietal sulcus (cIPS; − 42, − 40, 50), and ipsilesional IPS (iIPS; 42, − 40, 50). The 12 cortical ROIs were not overlapped with any subcortical lesions. Hence, we didn’t remove lesion voxels from individual ROIs. The averaged BOLD signals of all voxels in each ROI were extracted and ROI pairwise associations were calculated using Pearson’s correlation coefficients (r; 66 pairs in total among 12 ROIs). The FC strength between each ROI pair was then calculated as the transformed r-values (i.e. z-scores) using Fisher r-to-z transformation. The FC between ROIs was expressed as “FC_ROI-ROI” and FC changes as “△FC_ROI-ROI”. ROI pairs and their anatomical locations were visualized by means of BrainNet Viewer 1.7 (https://www.nitrc.org/projects/bnv/).