2026 Volume 53 Issue 3 Pages 175-182
Objective: To present and validate a CT-based spatial normalization framework for brain imaging that supports lesion-based outcome prediction and group-level analyses in stroke rehabilitation.
Methods: We retrospectively analyzed data from 25 patients with unilateral supratentorial intracerebral hemorrhage who were admitted to a convalescent rehabilitation ward. The hemorrhagic lesion on each patient’s CT image was segmented and converted to a binary mask. The whole-brain CT image and the lesion mask were spatially normalized onto an age-specific CT template using the Clinical Toolbox for Statistical Parametric Mapping. Voxel-based lesion–symptom mapping (VLSM) was used to relate lesion location to maximal gait speed at discharge.
Results: Spatial normalization was successful in all patients and yielded accurate anatomical alignment despite variability in routine clinical CT acquisition. VLSM identified a cluster in the posterior part of the posterior limb of the internal capsule. Overlay with a white matter atlas indicated that this cluster corresponded to corticospinal tract fibers innervating the lower limb muscles, consistent with previous MRI-based reports.
Conclusion: VLSM using spatially normalized CT images identified brain regions consistent with those reported in previous MRI-based studies, thereby demonstrating the validity of this method. This approach may facilitate wider implementation of quantitative neuroimaging in general hospitals.
