Work

Lesion-specific EEG signatures in stroke: a multi-scale framework integrating oscillations, connectivity, and asymmetry for machine learning decoding

Year: 2026

Type: article

Abstract: BACKGROUND: Lesion location is a major source of post-stroke neurophysiological heterogeneity, yet most electroencephalography (EEG) studies analyze patients as a single group, limiting lesion-specific biomarkers and translation. We proposed a lesion-centric, multi-scale EEG framework integrating local oscillations, inter-regional connectivity, and hemispheric asymmetry with machine learning to characterize and decode basal ganglia (P1), fronto-temporal/centrum semiovale (P2), and brainstem (P3) lesions. METHODS: Five-minute eyes-open, 128-channel resting EEG ([Formula: see text]) was recorded in 57 subacute stroke patients (P1 = 22, P2 = 18, P3 = 17) and 22 matched controls. From artifact-minimized [Formula: see text] segments, ROI-averaged power spectral density (PSD) (θ: 4–[Formula: see text]; α: 7–[Formula: see text]; [Formula: see text]: 12–[Formula: see text]; peak α frequency), current source density (CSD)-based magnitude-squared coherence, and directional BSI (dirBSI) were computed. Between-group and subgroup differences were assessed using t-tests/Wilcoxon and ANOVA/Kruskal–Wallis with Benjamini–Hochberg FDR ([Formula: see text]). EEG–behavior associations were examined with Spearman correlations. For machine learning, common spatial patterns (CSP) features were classified using linear discriminant analysis (LDA) with leave-one-subject-out cross-validation. To align with clinical workflow, we report HC vs P as “stroke detection/screening” and patient-only P1/P2/P3 classification as “lesion subtype decoding for stratification” (along with pairwise P1 vs P2, P1 vs P3, and P2 vs P3 models). An EEGNet baseline was evaluated for comparison. RESULTS: Increased α power and a leftward peak α shift were observed in patients (HC: [Formula: see text]; P: [Formula: see text]; [Formula: see text]). Pre-FDR, θ-band frontal–motor connectivity was strengthened, while posterior P–O connectivity in α/[Formula: see text] was weakened. Ipsilesional dominance in θ was indicated by dirBSI (HC: [Formula: see text]; P: [Formula: see text]; [Formula: see text]). Across lesions, [Formula: see text] power differences in central/parietal/occipital ROIs were detected pre-FDR, with higher parietal [Formula: see text] in P3; α-band asymmetry was stronger in P1/P2 and more symmetric in P3 ([Formula: see text]). EEG–behavior correlations did not survive FDR. Using CSP+LDA, accuracies of 92.41% (HC vs P), 94.87% (P1 vs P3), 85.71% (P2 vs P3), and 82.50% (P1 vs P2) were achieved; all binary AUCs exceeded 0.85; three-class accuracy reached 85.96%. CONCLUSION: This multi-scale EEG framework identifies lesion-associated neurophysiological signatures and demonstrates feasible lesion subtype decoding, supporting the potential of EEG biomarkers for objective stratification and precision neurorehabilitation.

Source: Journal of NeuroEngineering and Rehabilitation

DOI: 10.1186/s12984-026-01947-2