Authors :
Presenting Author: Levon Papikyan, BS – Yerevan State University
Meghana Chamarty, BS – Yale University
Hitten Zaveri, PhD – Yale University
Aline Herlopian, MD – Yale University
Rationale:
Pathological high-frequency oscillations (HFO) are biomarkers of epilepsy. Emerging data suggest that physiological and pathological HFO organize into clusters and are not static electrophysiological biomarkers(1, 2). They display dynamic changes during various phases, including the immediate pre-ictal phase, medication adjustments, and interictal sleep and non-sleep phases.
Methods:
5 patients with refractory frontal and temporal epilepsy undergoing intracranial monitoring were retrospectively analyzed. We analyzed 30-minute epochs from each patient selected, 4-6 hours farthest from the seizure and during different periods: immediate post-operative, the lowest medicationpreceding spontaneous seizure emergence, motor examination (only awake state), the highest medication following spontaneous seizures, post-operative mapping at conclusion of study, and 1Hz and 50Hz stimulation eliciting triggered seizures. We utilized the MNI HFO detector to identify HFO in ripple and fast ripple ranges. These HFO events were later spatially clustered using the Yale Brain Atlas(3). The clusters were further analyzed using a spatial-temporal density-based algorithm and multi-step sequence analysis to identify consistent patterns of occurrence (i.e. sequences) across the HFO events.
Results:
We identified two HFO clusters with significant differences in event length/duration (milliseconds), mean frequency of event (in Hz), and Root Mean Square (RMS, representing change in amplitude compared to baseline) values across clusters (Figure 1). Comparative statistical tests (Kolmogorov-Smirnov, t-test, and F-test) confirmed that the differences in HFO characteristics between clusters were statistically significant (p < 0.05).