Abstracts

Rationale: High frequency oscillations (HFOs) (80-600 Hz), which include ripples (80-250 Hz) and fast ripples (250-600 Hz), have been recently studied as potential biomarkers for localization of epileptogenic zones. However, one barrier for clinical application of HFOs is the separation of pathological and physiological HFOs. The objective of the present study is to use noninvasive technologies and normative database to distinguish pathological from physiological HFOs for localizing epileptogenic zones for pediatric epilepsy. Methods: HFOs in 30 children with medically intractable epilepsy and 60 healthy controls were studied using magnetoencephalography (MEG) and scalp electroencephalography (EEG). HFOs were automatically detected with a Stacked Sparse Autoencoder-based detector, one type of intelligent algorithms, and then visually verified with grid virtual sensors (VS) at source levels. The strength of HFOs was quantified with accumulated source imaging (ASI). HFOs in healthy controls were considered as physiological HFOs. Pathological HFOs were identified by measuring the peak amplitude of HFOs at virtual sensors. Epileptic areas were delineated by subtracting physiological HFO ASI from pathological HFO ASI. The sources of pathological HFOs in epilepsy were compared to clinical seizure onset zone determined by invasive recordings (current “gold standard”). Results: Compared to HFOs in controls in the same brain areas, HFOs in epilepsy were significantly enhanced in amplitude (239 +/- 147 vs. 46 +/- 21, p < 0.001). In addition, HFOs in epilepsy appeared to be prolonged burst with stereotype waveform patterns. Accumulated sources of HFOs in controls were predominantly in the deep brain regions, precuneus, and middle prefrontal regions while accumulated sources of HFOs in epilepsy showed sources in additional cerebral cortices. Sources of HFOs in epilepsy without removing physiological HFOs were overlapping with seizure onset zones in 19 patients (19/30, 63%) while sources of pathological HFOs (after removing physiological HFOs) in epilepsy were overlapping with seizure onset zones in 26 patients (26/30, 87%). Conclusions: MEG/EEG normative data might provide a noninvasive way to distinguish pathological from physiological HFOs for clinical applications of HFOs in pediatric epilepsy. Identification of pathological HFOs could facilitate the localization of epileptogenic zones for presurgical evaluation of epilepsy surgery. Funding: The project described was supported by Grant Number R21 NS104459 from the National Institute of Neurological Disorders and Stroke (NINDS), the National Institutes of Health. The normative database used in the present study was partially supported by Grant Number R21NS081420 and R21NS072817 from NINDS.