Article révisé par les pairs
Résumé : Objective: To compare and integrate hemispheric lateralization and spatial patterns of language mapping derived from functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG). Method: Twenty right-handed healthy adults performed three language tasks (verb/noun generation, phonological/semantic fluency, and sentence completion). A unified framework ensured methodological consistency across modalities: (i) identical tasks; (ii) individual mapping of block-level signal changes in fMRI and broadband (4–40 Hz) oscillatory power decreases in MEG; (iii) subject-level spatial extent-based thresholding of intramodality maps and quantification of intermodality overlap. Laterality indices (LI) were compared and combined across tasks and modalities to assess lateralization discordances. Spatial patterns were obtained by averaging individual maps, while permutation statistics evaluated intermodality spatial differences. Results: Task- and modality-combined LI reduced intermodality lateralization discordance from 45% to 10%. MEG showed twice the LI variability of fMRI. Overlap was concentrated in left fronto-temporal regions, while spatial differences emerged in left fronto-parietal areas (fMRI) and temporal/opercular regions (MEG). Conclusion: Although core language areas were identified in both modalities, modality- and task-integration may enhance robustness of language mapping in healthy subjects. Significance: Lateralization discordances and modality-specific spatial patterns, along with MEG's higher lateralization variability, suggest that multimodal integration could aid in presurgical clinical decision-making.

Documents en relation

DI-fusion