Abstracts

Rationale: High-density EEG (HD-EEG) has gained a lot of attention in clinical research and practice in recent decades. It has been generally accepted that the increase of recording channels could also increase the localization accuracy (Lantz et al., 2003, Song et al., 2018). With better spatial coverage, HD-EEG could potentially detect the neural activities in the deep sub-cortical structures (Seeber et al., 2019). However, it is very impractical to review HD-EEG visually. Automated spike detection algorithm has been proposed by our group previously (Song et al., 2016) to facilitate the review. In this pilot study, in order to validate the HD-EEG spike detection and source localization, we performed simultaneous intracranial EEG and scalp EEG recordings. At this moment, due to the concern of infection, we were not able to record intracranial EEG and HD-EEG concurrently. Therefore, we will record the scalp EEG with a limited number of electrodes. Methods: Seven focal epilepsy patients (> 18 years old) were enrolled in this study. A ~40-min HD-EEG (256 channels) were recorded with eyes closed using the Geodesic EEG 400 system. Patients were instructed to be sleep-deprived before the recording session in order to provoke more interictal spikes. The HD-EEG data were pre-processed by tools offered in NetStation and Matlab programs developed in-house. The interictal spikes were automatically detected and classified using our own Matlab programs (Song et al., 2016), and visually inspected and confirmed by epileptologists. The dipole locations at the peak location of each spike and averaged spike were calculated using a three-layer realistic head model in Brainstorm. In order to validate the spike detection and localization, we were able to add a limited number (n=~12) of scalp electrodes on these patients during their SEEG monitorings. The limited scalp EEG montage was tailored to each patient during the pre-surgical planning stage: 1) to avoid infection, and 2) to be able to capture the “phase-reversal” feature of the spike on the scalp based on pre-operative HD-EEG. The limited scalp EEG electrodes were placed and wrapped on the patient’s head in the sterile field by experienced EEG technologists post-operatively. During the post-operative recording stage, the spikes were marked by epileptologists manually in the SEEG recordings. Segments of the scalp EEG time-locked to the SEEG spikes were averaged and compared to the pre-operative HD-EEG spikes to see if they shared the same scalp representation features. Results: Among these seven patients, three of them had extra-temporal resections, two of which had previous resections. Four patients had mesial temporal sclerosis (MTS). The HD-EEG spike localizations of these patients were within close proximity of the seizure onsets identified by SEEG. Spikes in the SEEG recordings were more frequent than those identified in the pre-operative HD-EEG recordings. The averaged scalp responses were very similar to the pre-operative HD-EEG recordings with respect to the morphology and relative amplitude. However, we were not able to differentiate the limited scalp responses from two independent spike sources in one patient, which were approximately 6 mm apart along the hippocampus. Interestingly, the pre-operative HD-EEG was able to distinguish these two independent events topographically, which, again, demonstrated the advantage of using high-density spatial coverage in HD-EEG. Conclusions: In this pilot study, we used simultaneous limited scalp EEG and SEEG to validate the HD-EEG spike detection and source localization. We found a high concordance between the simultaneous recordings and pre-operative HD-EEG findings regarding the spike’s source location, morphology, and relative amplitude. Funding: No funding