Abstracts

Rationale: Evaluation of language is a critical step prior to the surgery. In clinical settings, there is no consensus on an optimal methodology for lateralizing language dominance. We propose a beta desynchrony approach based on dynamic imaging of coherent sources (DICS) analysis of task magnetoencephalography (MEG) to accurately localize and lateralize language production network in patients with focal epilepsy. Methods: We examined MEG data from 29 patients (age 42±13 years, 14 females) completing an auditory definition naming (DFN) and a visual picture naming (PN) language experiments. Patients overtly named the auditorily and visually presented common nouns during DFN and PN, respectively (Fig.1A). Source model was constructed using a regular 3D grid of individual’s MR with 8mm resolution in MNI space. Sources were estimated using a DICS beamformer in a beta frequency range of 18-25 Hz and post-stimulus time intervals of 1100-1700ms and 500-1100ms corresponding to DFN and PN, respectively, against a pre-stimulus time interval of -300-0ms. Time intervals were selected based on the peak desynchrony effects in time-frequency sensor-level data and conditioned to be at least 100ms prior to average speech onset responses (Fig. 1B). Cluster-based permutation testing was applied for within-subject (a paired t-test of pre- and post-stimulus activity) and between-subject (an independent sample t-test against a null hypothesis) statistical inferences. Atlas-driven laterality indices (LIs) were computed for 21 language-specific frontotemporal cortical regions Results:

DFN and PN group localization consistently revealed beta desynchrony effects in canonical perisylvian language areas with an excellent overall correlation of r=0.89 (q<0.05) within language-specific brain regions (Fig. 1C). Comparing the localization maps, DFN showed stronger effects in the left prefrontal cortex (left inferior frontal gyrus, pars opercularis t=5.6 and pars triangularis, t=5.3) whereas PN showed widespread power desynchrony activation in left temporal, parietal and frontal cortices (supramarginal gyrus, t=5.2, superior temporal gyrus t=4.9, precentral gyrus t =4.8, and IFG pars triangularis, t=4.7) as well as right precentral gyrus, t = 4.6. Laterality analysis consistently showed a left-dominance (LI > 0.25) for most individuals (82% in DFN and 93% during PN), Fig. 1D, with average LI indices of 0.40+-0.25 and 0.34+-0.20 for DFN and PN, respectively (Fig. 1E).

 Conclusions: Findings consistently demonstrated strong left-hemispheric dominance of language production network in patients with focal epilepsy. The proposed approach is robust in determining the language dominance in a heterogeneous population of neurosurgical candidates and thus it is recommended for clinical examinations. Funding: Supported by The Medical College of Wisconsin SRA