Poster F105, Tuesday, March 28, 8:00 – 10:00 am, Pacific Concourse
Language lateralization assessed by magnetoencephalography imaging using three different language tasks
Elke De Witte1, Leighton Hinkley1, Danielle Mizuiri1, Coleman Garrett1, Susanne Honma1, Heidi Kirsch1, John Houde1, Mitchel Berger1, Sri Nagarajan1; 1University of California, San Francisco
Introduction: Magnetoencephalography (MEG) is a valuable alternative to the Wada test for defining preoperative language lateralization in neurosurgical candidates. The goal of this study was to benchmark an existing preoperative MEG paradigm for language lateralization with excellent predictive value (verb generation) against new language tasks. Methods: In this study, a small cohort of 7 glioma patients underwent both the Wada test and MEG using a 275-channel whole-head biomagnetometer (CTF Systems) for verb generation (VG), non-word repetition (NWR) and picture naming (PN) in the preoperative phase. In a large cohort of 67 and 59 glioma patients MEG was conducted using respectively VG + NWR and VG + PN. MEG data was reconstructed in source space using an adaptive spatial filtering technique in Nutmeg (www.nitrc.org/projects/nutmeg). Optimized time windows and regions of interest for each task were determined by examining oscillations in the beta band (12-30Hz), which were concordant with the Wada data or VG laterality index (LI). Results: The following time windows showed the highest correlation with VG LI or the WADA data: -450 -250ms for PN response locked; -350 -150ms for NWR response locked and 650 850ms for NWR stimulus locked. Using these time windows we identified 83.34%, 87.5% and 100% concordance with the Wada test for respectively NWR, PN and the composite LI for all 3 language tasks. 80% of the large cohort data showed concordance with the VG LI. Conclusion: This study enables to define the optimal parameters for MEG laterality calculations in language tasks outside of verb generation.
Topic Area: METHODS: Neuroimaging