Patients underwent a clinically indicated brain 3-T MRI with contrast as part of their stroke care. All sequences included an Axial T1, Axial Diffusion-Weighted Image (DWI), Axial T2* GRE, and Axial T2 FLAIR. Infarct volumes were calculated by manually tracing hyperintense areas of diffusion restriction on patient T2 DWIs using MRIcron© (NeuroImaging Tools & Resources Collaboratory) by a reviewer masked to patient characteristics and stroke subtypes. Such techniques, collectively known as voxel-based morphometry, have been well-validated for calculating infarct volumes in similar research [12 (link), 13 (link)]. 30% of the MRI traces were re-reviewed by another reviewer, masked to patient characteristics, to determine interrater reliability, which was greater than 90%.
All scans were normalized to a reference scan to standardize voxel size, number of axial slices, and resolution. Volumes were calculated using Statistical Parametric Mapping 12 (SPM12; Wellcome Center for Human Neuroimaging, London, UK), which operates using MATLAB R2007a (7.4) to R2019b (9.7) (Mathworks, Natick, MA, USA). For analysis, all infarct volumes were converted from mm3 to milliliters (mL), and a 5 mL increment was used when reporting results.
All scans were normalized to a reference scan to standardize voxel size, number of axial slices, and resolution. Volumes were calculated using Statistical Parametric Mapping 12 (SPM12; Wellcome Center for Human Neuroimaging, London, UK), which operates using MATLAB R2007a (7.4) to R2019b (9.7) (Mathworks, Natick, MA, USA). For analysis, all infarct volumes were converted from mm3 to milliliters (mL), and a 5 mL increment was used when reporting results.