Abstracts

Rationale: Epilepsy remains challenging to treat, with about one-third of patients unresponsive to medications. Surgical removal of pathological tissues can alleviate seizures if the epileptogenic zone (EZ) is accurately identified. High-frequency oscillations (HFOs) are promising biomarkers for locating the EZ and guiding surgery, though their identification and clinical use have shown variable results. This study focuses on spatiotemporally associated HFO sequences to identify pathological HFOs and correlate them with the EZ in focal epilepsy patients. By examining HFO organization and stability, the study aims to understand their dynamics and relation to the seizure onset zone (SOZ), enhancing surgical targeting precision.

Methods: We collected and analyzed pre-surgical MRI, CT, and iEEG recordings from 40 patients with medically intractable epilepsy from two clinical centers. Individual head models and iEEG electrode positions were modeled and co-registered, and approximately 24 hours of iEEG recordings were examined for each patient. Our developed algorithm automatically identified over 8 million potential HFOs (including pathological and physiological, aHFOs) present in the iEEG recordings. We then extracted and screened the detected events for multichannel HFO sequences (HFO-seq), defined as a series of HFO events occurring across multiple electrodes within a short period, using our proposed approach. We investigated the spatial distribution of all occurring HFOs and HFO-seq detected for each patient and compared them to the clinically determined SOZ. Furthermore, we examined the temporal and spatial relationships between individual HFO events within sequences across the subject group.

Results: Over 8 million putative HFOs (53% ± 13.74% during the day, n=40) were detected, and approximately 274,000 HFO sequences (48% ± 15.84% during the day) were identified. Our analysis revealed a significantly higher occurrence rate asymmetry for HFO-seq towards the SOZ compared to all HFOs (aHFOs: 0.39 ± 0.33, HFO-seq: 0.89 ± 0.13; p< 10^-7, two-sided Wilcoxon signed-rank test). We observed consistent temporal relationships and unique spatial patterns across cortical areas and identified channels with high HFO recruitment, overlapping with the SOZ, especially for patients with seizure-free outcomes. Temporal recruitment of spatial regions varied across sequences, showing heterogeneous onset and progression patterns. Furthermore, high repeatability of single sequences was found, with significant sequence similarity (0.43 ± 0.07) surpassing a shuffled control group (p< 10^-7). These findings suggest that HFO sequences involve unique, repeatable, and stable patterns of spatiotemporal propagation across cortical locations, providing insights into epileptogenic network dynamics.



Conclusions: This study developed an approach to identify pathological HFO sequences in focal epilepsy patients. These sequences, closely associated with the clinical EZ, exhibited unique, heterogeneous, repeatable spatiotemporal dynamics, offering insights into epileptogenic networks and potentially improving surgical interventions.

Funding: This work was supported in part by NIH R01 NS096761 and EB021027.