Abstracts

Rationale: The responsive neurostimulation (RNS) system has emerged as an effective adjunctive therapy in medically intractable epilepsy [Neurology. 2011 Sep 27;77(13):1295-304]. RNS data offers a unique opportunity to follow electrocortical activity (ECoG) over extended time windows with high signal integrity from direct, chronic cortical implantation. Continued study of RNS data can yield new understandings of ictal-interictal activity, diurnal and seasonal changes, treatment responses to RNS and medications, and the physiology of neural oscillations. Frequency analysis techniques for detecting early ictal changes are limited, but clinical applications for EEG frequency analysis are expanding [Seizure. 2018 Feb;55:70-75]. The purpose of this study was to develop frequency analysis techniques based on specific ictal and interictal patterns that could be consistently identified in RNS ECoG samples. Methods: A database comprised of RNS ECoG recordings from 15 RNS-implanted patients at the University of Miami Epilepsy Center was utilized to identify novel, consistent ictal and interictal patterns. The patient population contained varying sites of seizure focus and electrode implantation: 6 bilateral mesial temporal; 4 unilateral temporal (mesial and lateral); 3 unilateral frontal and temporal; 1 mesial temporal and occipital; 1 parietal and occipital. The implanted RNS device acquires 90 second ECoG samples at scheduled times (e.g. 0200 and 1300) daily, and during prolonged ictal periods where treatment from the device is usually triggered -recording device treatments and response. We conducted a spectrographic analysis of ECoG samples to characterize high frequency activity (25 - 125 Hz), low frequency activity (0 - 25 Hz), and highly active narrow frequency bands (5 Hz bandwidth) within ictal recordings and scheduled recordings that were classified as either interictal with active discharges or interictal without active discharges. 25 samples for each type of recording (ictal, active discharges, no discharges) were analyzed for each patient, totaling 1125 ECoG recordings reviewed. Consistent patterns within high frequency activity, low frequency activity, and active narrow frequency bands were summarized for each patient and compared across the database. Results: Reductions in interictal high frequency activity; development of a high power, narrow, 5-10 Hz band; and the disappearance of narrow, higher frequency bands seen in interictal recordings were observed to correlate with the onset of ictal activity in every seizure. These findings did not vary among location of electrode implantation or epileptogenic foci. Conclusions: Our findings suggest patterns of pre-ictal changes in recorded ECoG samples that occur independent of the location of ictal onset. These patterns could be further investigated to be used as potential markers, or early indicators of seizure onset to improve RNS detection and treatment response. Funding: No funding