Abstracts

Rationale: Magnetoencephalography (MEG) and high-density electroencephalography (hd-EEG) are useful in presurgical evaluation of patients with drug-refractory epilepsy, as they improve the localization of the epileptic focus, surgical decision and potentially, post-surgical outcome. Nonetheless, these two techniques are underused as they are costly, time-demanding, and difficult to handle in everyday clinical practice. In this study we evaluated the localization performance of an ‘optimal’ EEG montage consisting in adding to standard 10-20 positions a few electrodes around the region where interictal epileptiform discharges show maximal amplitude. This would be a reasonable strategy under the assumption that the topographical distribution of the electric field in the vicinity of the focus is more relevant to source localization.

Methods: We retrospectively analyzed data of patients who underwent simultaneous EEG-MEG recordings (24 patients) and for whom the epileptic focus could be localized thanks to invasive EEG or epilepsy surgery (gold-standard). We compared the following strategies for source localization of interictal activity: a) MEG source localization, b) EEG-MEG fusion source localization, c) low-resolution EEG source localization (21 channels), d) ‘optimal’ EEG source localization (10-20 + n channels around the peak of the spike), e) same number of electrodes as (d) but randomly drawn from a 64 10-10 layout. The primary endpoint was the distance of the peak of the source from the epileptic focus.

Results: Best localization performance was achieved by MEG and EEG-MEG fusion (localization error 7.4mm and 7.5 mm). Low-resolution EEG had the worst performance (16mm localization error), but adding a few channels optimally located quickly improved localization performance (22ch = 14mm, 23 channels = 13 mm, 24 channels = 8 mm) (Figure 1). Adding more channels did not further improve the performance. Drawing additional channels randomly did not improve the localization error.

Conclusions: source localization from optimal EEG montage with just 24 channels achieves a localization error roughly comparable to MEG and EEG-MEG fusion, while being cheaper, easier, time-saving. Optimal EEG source localization could be explored in clinical setting whenever hd-EEG and MEG are not readily available.

Funding: Please list any funding that was received in support of this abstract.: Frederick Anderman Fellowship in Epileptology and EEG (McGill University) and GR-2019-12368960 from Italian Ministry of Health to GP.