Differential roles of DTI and EEG in predicting cognitive function after left basal ganglia stroke: a...

Diffusion tensor imaging (DTI) and electroencephalography (EEG) may serve as structural and functional biomarkers for cognitive impairment following basal ganglia stroke; however, their relative contributions are uncertain. This study compared findings of DTI and EEG between healthy controls and patients with left basal ganglia stroke without cortical lesions and identified cognitive outcome imaging predictors. Twenty-seven participants (17 controls, 10 patients) underwent comprehensive cognitive testing, DTI, and EEG. Two DTI metrics (fractional anisotropy [FA], mean diffusivity [MD]) were extracted from major association tracts involved in cognitive processing. EEG indices included the alpha-to-beta ratio and delta-to-alpha ratio across cortical lobes. Group comparisons demonstrated significant global cognitive decline in the stroke group. Structural alterations included reduced FA in ipsilesional association tracts and increased MD in ipsi- and contralesional tracts. No significant group differences were observed in EEG spectral indices. Brain-behavior associations were examined using covariate-adjusted partial correlations with Benjamini-Hochberg false discovery rate (FDR) correction. No DTI-cognition associations survived FDR correction in controls, and only four DTI association remained significant in pooled stroke analyses. In contrast, 25 EEG-cognition associations survived FDR correction in pooled stroke analyses, whereas none survived in controls. Higher ABR values were associated with better cognitive performance, while higher DAR values were associated with poorer cognition. These findings suggest that DTI and EEG provide complementary information, with structural metrics reflecting white-matter integrity and EEG indices demonstrating more widespread associations with cognitive performance following subcortical injury. Functional EEG measures may offer sensitive markers of post-stroke cognitive vulnerability when interpreted alongside structural imaging. As a proof-of-concept study in a small, strictly selected cohort, these findings should be regarded as preliminary and hypothesis-generating, requiring replication in larger samples.