T161. Correlating magnetoencephalography (MEG) and high-density electroencephalography (EEG) with invasive recordings in pediatric patients undergoing epilepsy surgery

Abstract

Introduction Epilepsy surgery is a potentially curative treatment for medically refractory epilepsy (MRE). The goal of epilepsy surgery is the resection of the epileptogenic zone. Before surgery, MEG and scalp EEG are used to estimate the epileptogenic zone noninvasively by localizing the irritative zone (IZ), which is the area where interictal epileptic discharges (IEDs) occur. Previous studies correlated MEG, scalp EEG and the gold standard intracranial EEG (iEEG), but no study compared their ability to localize IEDs in the same cohort solving the inverse problem. Our goal is to assess the ability of 306-sensors MEG and 70-channels EEG (HD-EEG) to localize the IZ in pediatric patients with MRE by comparing them with iEEG findings. Methods We analyzed interictal data from 25 children with MRE who underwent iEEG. We analyzed 12 min of simultaneous MEG and HD-EEG data, and 10–15 min of iEEG data recorded with subdural and/or depth electrodes. IEDs were detected on MEG, HD-EEG and iEEG data by two reviewers who were blinded to each other’s results. We determined the location of each iEEG contact on the patient’s cortex by co-registering presurgical MRI with post-implantation computerized tomography. All IEDs were localized using an Equivalent Current Dipole (ECD) method (forward model: Boundary Element Method). Only ECDs with goodness-of-fit >75% were considered. We measured the minimum distance of each MEG and HD-EEG ECD from iEEG ECDs. We classified each HD-EEG and MEG ECD as: (i) not-covered, if in a brain area not covered by iEEG (>30 mm far from iEEG); (ii) concordant, if within iEEG coverage and close to an iEEG ECD ( We also calculated, for each patient, precision and sensitivity of MEG and HD-EEG, as well as their localization error (i.e. the mean distance of their ECDs from iEEG ECDs). Results We included 13 patients in the analysis. HD-EEG showed a higher proportion of ECDs outside the iEEG coverage than MEG (26% vs 5%). MEG showed a higher proportion of ECDs concordant with iEEG compared to HD-EEG. MEG presented a higher precision to the iEEG ECDs than HD-EEG (50% vs. 31%, p = 0.04), with no difference in sensitivity. The localization error of MEG was lower than HD-EEG (p = 0.017). In 7 patients, MEG ECDs were closer to iEEG ECDs compared to HD-EEG (p Conclusion We compared for the first time the source localization of three non-invasive and invasive neuroimaging methods (HD-EEG, MEG, and iEEG) in the presurgical evaluation of children with MRE. Our findings showed that MEG can localize IEDs with lower localization error and higher precision compared to scalp EEG, even though EEG is more accurate than MEG in few cases. MEG and HD-EEG are complimentary valuable tools in the non-invasive mapping of IZ in epilepsy surgery.