Abstract
BACKGROUND: Ischemic stroke disrupts structural connectivity between grey matter regions, influencing functional outcomes. Diffusion MRI-based tractography allows assessment of white matter damage and may guide rehabilitation. It remains unclear how structural connectivity disruption (SCD) estimated from individual tractography compares with model-based estimates using normative tractograms. METHODS: Diffusion MRI data were analyzed from 23 individuals with subacute ischemic stroke and 60 healthy adults from two age-specific cohorts (Human Connectome Project/HCP, Alzheimer's Disease Neuroimaging Initiative/ADNI3). In Experiment 1, SCD estimates from individual tractography were compared to scores from the model-based Network Modification (NeMo) tool in the stroke cohort. In Experiment 2, synthetic middle cerebral artery lesions were applied to healthy adults' diffusion data to compare SCD estimates from tractography on synthetically lesioned versus complete data, to SCD estimates from the model-based NeMo tool. Rank-based linear mixed-effects models compared differences across approaches, cohorts, and 13 regions of interest (ROIs). RESULTS: In stroke, SCD scores from individual tractography and NeMo were strongly correlated (R(2) ≥ 0.81), with largest differences in the inferior parietal gyrus (median [IQR]: -4.2% [-9.2-1.4]). In healthy controls, NeMo underestimated SCD relative to individual tractography, with larger median SCD differences across ROIs in older (11.8% [7.7-13.3]) than younger adults (-6.7% [-9.3--1.9]). Cohort and ROI explained 47-69% of variance in SCD difference across approaches. CONCLUSION: Both individual tractography and the model-based framework capture regional SCD patterns, with larger differences in older adults. The model-based framework is efficient for large-scale studies, while individual tractography may better capture patient- and age-specific SCD.