Abstracts

Rationale: Source localization of inter-ictal spike activity detected on magnetoencephalography (MEG) is an established method in the pre-surgical evaluation of epilepsy patients for epileptogenic zone identification. However, the significance of other focal inter-ictal patterns such as high-frequency oscillations is less clear. Focal beta (12?"30 Hz) activity can often be seen on electroencephalography (EEG) and MEG, and in this preliminary study, we examined the feasibility and utility of localizing MEG focal beta oscillations in a cohort of individuals with focal epilepsy. Methods: MEG studies obtained as part of the pre-surgical epilepsy work-up were retrospectively reviewed for documentation of focal beta oscillations. Four-second epochs around each event, taken from the 40?"60 minutes of recording for each patient, were generated using command line tools from CTF MEG (v5). Interference was removed using a dual signal subspace projection (DSSP) algorithm (Sekihara et al., 2016). Source localization was performed using the Champagne algorithm (Owen et al., 2011) in the NUTMEG toolbox (Neurodynamic Utility Toolbox for Magnetoencephalography, v4.6)(Dalal et al., 2011). The active time window was the duration of the beta activity, and the control time window was an equal duration ending 100 ms before the start of beta activity. Localization of the signal of interest was compared with EEG, MRI, MEG dipole modeling, and intraoperative electrocorticography (ECoG) obtained during the pre-surgical evaluation. Concordance was defined as within 2 cm or a similar approximate distance when the study modality did not allow direct measurement. Post-operative outcomes were assessed as to whether the beta activity was part of the surgical intervention. Results: Ten individuals with focal beta activity were included, and source localization resulted in reliable results in nine individuals. Subject characteristics are reported in Table 1. In the three subjects with non-lesional imaging, localization was concordant with other modalities. In both non-lesional subjects with a good interventional outcome, the intervention included the region identified by beta localization. In the third, beta localization identified a region encompassing the inferior and middle frontal gyri while the resection involved the superior frontal gyrus, and there was no improvement in seizure burden. Of four additional subjects without gross anatomic distortions, beta localization was concordant with other modalities in three. Interventional outcomes were good in two of these individuals, while the third subject had RNS implantation with ongoing programming at the time of last follow-up. In the fourth subject, the beta oscillations were localized to the left frontal lobe while other modalities identified the left temporal lobe, and anterior left temporal lobectomy resulted in a good outcome. Results of the beta localization were less reliable in three subjects with gross anatomic distortions due to tumor resection (1 subject; localization had concordant laterality but remote from the likely seizure focus) or holohemispheric atrophy (2 subjects; laterality discordant in one and parenchymal localization failed in the other). Conclusions: Beta oscillations can occur in or near epileptic tissue and can be detected and localized using advanced MEG source localization methods. This preliminary study suggests that in non-lesional cases, where there is the greatest need for improved identification of the seizure onset zone, MEG localization of beta oscillations could provide useful adjunctive information and deserves further study. Funding: Support provided by the following grants: R01DC13979, R21DC14525, R01DC10145, NSF BCS 1262297, U01DC013029, R21 NS076171.