Determining hemispheric language dominance from MEG beta-power modulations: Concordance with fMRI

Accurate identification of the language-dominant hemisphere is essential for surgical planning in drug-resistant focal epilepsy. Functional MRI is widely used as a noninvasive alternative to the Wada test but may be contraindicated or unreliable in some patients, motivating complementary approaches such as magnetoencephalography (MEG). We tested whether beta-band MEG power suppression can provide reliable language lateralization when combined with a task-matched nonlinguistic control, participant/ROI-specific timing, and a multi-threshold weighted-bootstrap laterality index (LI). Seventy-four adults with temporal lobe epilepsy completed visual semantic-decision and false-font control tasks during MEG. Beta-band (17-25 Hz) source power was estimated with DICS beamforming, and LIs were computed in frontal, temporal, angular, and composite lateral ROIs using 300-ms sliding windows. For comparison, participants also completed an auditory semantic-decision versus tone-decision fMRI protocol, with fMRI LIs derived from matched ROIs. MEG showed a left-lateralized spatiotemporal sequence in which frontal LI emerged earlier than temporal-parietal LI, following early bilateral temporo-parieto-occipital engagement. Compared with a prestimulus baseline, the false-font control contrast increased LI magnitude and improved MEG-fMRI correspondence, with a peak MEG-fMRI correlation of r = 0.65. Participant/ROI-specific timing produced only small and mixed changes in concordance. Using the optimized MEG pipeline, MEG and fMRI agreed in 67/74 patients (90.5%; 95% CI: 83-96%; κ = 0.81). Bootstrap LIs performed similarly to magnitude- and vertex-count-based variants while providing confidence intervals. These findings suggest that beta-band MEG suppression during semantic decision, especially when contrasted with a nonlinguistic control task, provides a temporally resolved and clinically feasible marker of hemispheric language dominance that is highly concordant with fMRI.