Authors :
Presenting Author: Paul Ferrari, PhD. – Helen DeVos Children's Hospital
Bahram Sarvi-Zargar, MS – Helen DeVos Children's Hospital; Erin Skrzypek, REEG T, CMEG – Helen DeVos Children's Hospital; Jeremy Gurumendi, REEG T, CMEG – Helen DeVos Children's Hospital; Angel Hernandez, MD – Helen DeVos Children's Hospital
Rationale:
Non-invasive language mapping is an essential part of a phase-1 presurgical workup in patients with refractory epilepsy. There remains large variation across magnetoencephalography (MEG) labs in terms of tasks and analysis methods for determining hemisphere dominance. One analysis approach that has shown to be effective is mapping frontal beta-band event-related desynchronization (ERD) during expressive language tasks, such as verb generation or picture naming. However, some patients struggle to comply with the task demands. In the present work we deployed a traditional auditory word receptive language task embedded with complex tone stimuli and analyze the beta ERD using a synthetic aperture magnetometry (SAM) differential beamformer contrasting the tone and word responses.
Methods:
MEG data from 34 refractory epilepsy patients (22 females, 1 left-handed, 3 ambidextrous) who had undergone a full battery of presurgical language mapping including the auditory tone-word receptive task were analyzed. The task stimuli consisted of 120 tone word pairs. The tone was an 800ms complex tone presented just prior to the word at a constant offset. Patients were instructed to remember four target words as they to listen to the tone-word pairs. Upon hearing a target word, patients overtly repeated that word. MEG data were epoched into four second trials centering on word onset. Only data from non-target words were analyzed, i.e. receptive listening. A SAM differential beamformer was used to map average volumetric beta-band ERD responses (15-30Hz) to word stimuli over the initial 600ms using the same time period of the tone activity as a baseline. The resulting volumetric beta maps were coregistered to the subjects’ T1 MRI brain image and warped to standard space for group analysis using non-parametric permutation statistics for identifying statistical significance. A region of interest (ROI) was selected based on group activation and virtual sensors for each patient were estimated using a LCMV beamformer at the ROI site and the homologous location in the contralateral hemisphere. ROI source activity was transformed into time-frequency response (TFR) maps from 1-50Hz to assess the power differences across the hemispheres for each patient.
Results:
The average group analysis revealed the strongest beta ERD response in left inferior frontal gyrus (IFG, BA44) (p< 0.05). The group ROI TFR maps show clear beta-band ERD in bilateral BA44 with a much stronger ERD in the left. Individual laterality assessment corresponded to the overall standard MEG battery 29 out of 34 cases (82%). The correspondence to fMRI laterality assessment was also 82%.