Abstracts

Rationale: A large proportion of pediatric patients with epilepsy is medically intractable and require resective surgery to remove the epileptogenic zone (EZ), usually estimated by the seizure onset zone (SOZ). High Frequency Oscillations (HFOs) (80-500 Hz) have become widely recognized in recent years as interictal biomarkers for the EZ [1], especially when concurrent with epileptic spikes [2]. However, they are not widely used during presurgical evaluation, and among the reasons is the difficulty to detect them noninvasively. The aim of this study was to use high-density scalp EEG to noninvasively detect interictal HFOs, and to distinguish between interictal spikes with and without HFOs, in order to investigate whether the spike-HFO concurrence represents an accurate interictal marker of the SOZ. Methods: We analyzed scalp EEG data from 5 children (8-15 years old) with medically refractory epilepsy. We developed an algorithm for the automatic detection of interictal HFOs and spikes on scalp EEG data. The algorithm distinguished real HFOs from artifacts of non-cerebral origin, by considering the time-frequency map as the ground truth: only the HFOs characterized by an isolated peak in the time-frequency plane (Fig. 1) were detected. Our algorithm also classifies interictal spikes as red spikes (RS), if overlapping with HFOs, or green spikes (GS), if not overlapping (Fig.2). We defined RS channels as the ones with a high-rate of RS, and GS channels as the ones with a high-rate of GS (Fig.3). The "high-rate" threshold was set for each patient using Kittler's and bootstrapping methods [3]. Each channel was also classified as inside or outside the SOZ, according to the SOZ identified by the epileptologist. Finally, we calculated Sensitivity, Specificity, and Accuracy to the SOZ of both RS and GS for each patient. The same measures were also calculated for all the spikes without considering HFOs. Results: We report results from 3 patients (in 2 patients seizures were generalized or interictal activity insufficient). RS showed the best performance in terms of specificity and accuracy to identify the SOZ in all 3 patients, compared to GS and to all spikes. The ranges of specificity and accuracy of the spikes increase, respectively, from 60-67 % to 75-90 % and from 61-67 % to 78 %, when considering the co-occurrence with HFOs and selecting only the RS. Conclusions: Our automatic detector represents a useful tool to noninvasively investigate the concurrence of interictal HFOs and spikes, which seems to provide an accurate marker of the SOZ. The possibility to detect HFOs noninvasively and automatically may foster the use of HFOs in the presurgical evaluation of patients with epilepsy, providing a new precise and noninvasive biomarker of the epileptogenic zone, which may improve patients' postsurgical outcome. 1. Jacobs J, Staba R, Asano E, Otsubo H, Wu JY, Zijlmans M, et al. High-frequency oscillations (HFOs) in clinical epilepsy. Prog Neurobiol 2012;98:302?"15. 2. Jacobs J, Kobayashi K, Gotman J. High-frequency changes during interictal spikes detected by time- frequency analysis. Clinical Neurophysiology. 2011; 122:32?"42. Funding: Epilepsy Foundation / American Epilepsy Society Research Grant: "Propagation of high frequency oscillations as biomarker of epilepsy" Harvard Medical School - OFD/RRRC/CTREC Shore Faculty Career Development Award: "Interictal High Frequency Oscillations as Biomarker of Epilepsy in Children"