Abstracts

Rationale: Although intracranially recorded interictal high-frequency oscillations (HFOs) have been proposed as a promising spatial biomarker of the epileptogenic zone, distinguishing pathological HFOs from physiological HFOs is challenging. We recently developed a deep learning (DL)-based algorithm to distinguish two types of HFOs: epileptogenic HFOs (eHFOs) and non-epileptogenic HFOs (non-eHFOs), in children with medication-resistant epilepsy who achieved seizure-freedom after resection. HFOs from the resected brain regions were initially labeled as eHFOs, and those from the preserved brain regions as non-eHFOs to train the model (Zhang et al. Under revision). We set out to quantify eHFOs and non-eHFOs in functionally different brain regions, based on the presence of spontaneous seizures and functional cortical mapping.

Methods: We retrospectively analyzed 19 consecutive children with medication-resistant neocortical epilepsy who underwent chronic intracranial EEG (iEEG) monitoring at the UCLA Medical Center. We defined four functionally different brain regions as: 1) the seizure onset zone (SOZ) (brain regions with spontaneous seizures during monitoring); 2) brain regions with stimulus-induced seizures or afterdischarges (ADs); 3) eloquent cortex (brain regions in which a patient exhibited behavioral changes with somatosensory or language disturbances with stimulation mapping); and 4) non-eloquent cortex (brain regions in which no seizures/ADs or behavioral response were elicited with stimulation mapping). A 90-minute recording of iEEG data during sleep was obtained from each patient, and our DL-based algorithm was applied to quantify eHFOs and non-eHFOs in each channel. The numbers of eHFOs and non-eHFOs were contrasted in each functionally different brain region.

Results: The analysis of 1,760 contacts across 19 patients yielded 88,499 HFOs (43,904 eHFOs [49.6%]) after artifact rejection. Cortical stimulation mapping was performed in 442 contacts across 18 patients. The rate of eHFOs was higher in the SOZ and brain regions with stimulus-induced seizures/ADs than in the eloquent cortex (median rate/ch/min: 0.28 vs. 0.11, and 0.17 vs. 0.11; p = 0.002, and p = 0.03 with Wilcoxon signed rank test, respectively) (Fig.1). There was higher ratio of eHFOs to non-eHFOs in the SOZ than in the eloquent cortex, and there was a trend towards higher ratio of eHFOs to non-eHFOs in brain regions with stimulation-induced seizures/ADs than in the eloquent cortex (median ratio: 1.22 vs. 0.73, and 1.68 vs. 0.73; p = 0.04, and p = 0.05 with Wilcoxon signed rank test, respectively).

Conclusions: We may be able to use eHFOs as a surrogate of pathological HFOs, and non-eHFOs as a surrogate of physiological HFOs. Quantification of both eHFOs and non-eHFOs may facilitate delineating the epileptogenic zone and the eloquent cortex more accurately.

Funding: Please list any funding that was received in support of this abstract.: The Susan Spencer Clinical Research Training Fellowship in Epilepsy from the American Academy of Neurology (#20184605), the Sudha Neelakantan & Venky Harinarayan Charitable Fund, the Elsie and Isaac Fogelman Endowment, and the UCLA Children’s Discovery and Innovation Institute (CDI) Junior Faculty Career Development Grant (#CDI-SEED-010121).