Abstracts

Rationale: High frequency oscillations are observed more frequently, and in closer proximity to the seizure onset zone than outside of it, suggesting that they may help identify epileptogenic networks. However, their temporal and spatial characteristics may vary by seizure type and between individuals. We hypothesize that the patterns of propagation of high frequency activity among cortical networks will help identify the epileptogenic zone, which could ultimately improve seizure localization for epilepsy surgery planning. Methods: Intracranial EEG recordings (ECoG) of six patients with focal and regional epilepsy were analyzed with a matching pursuit (MP) method to identify ECoG signals containing high frequency ictal activity. Signals were parameterized using multivariate autoregressive models (MVAR), and analyzed with event-related causality method (ERC). ERC estimates the direction, intensity, and temporal evolution of neural activity propagation among cortical sites, and statistically compares non-stationary time series of interest (e.g. seizure) with non-stationary reference time series (e.g. interictal interval). We used ERC to study the patterns of propagation of high frequency activity (70-170 Hz) during inter-, pre-, and ictal intervals, as well as significant differences between the patterns of inter-, pre-, and ictal activity propagation. Results: In patients with focal epilepsy, we observed a common sequence of increases in high-frequency propagation when ictal intervals were compared with interictal or pre-ictal intervals. Immediately after seizure onset, propagations occurred from sites identified by epileptologists as ictal onset zones to all other analyzed recording sites. Next, ictal propagations were observed between ictal onset zones and sites in close proximity to these onset zones. Interestingly, subsequent increases of ictal propagation also involved sites inferred as onset zones from pre-surgical scalp recordings. Changes in high-frequency propagation between the ictal onset zone and all other analyzed recording sites were also observed in the post-ictal interval (compared to inter- or preictal intervals). In patients with poorly localized (regional) ictal onsets, these patterns were not observed. Conclusions: Comparisons of patterns of high-frequency activity propagation during inter-, pre-, ictal, and post-ictal intervals may be useful for localizing the epileptogenic zone for epilepsy surgery. The seizure onset zone may be characterized by increases in high frequency activity propagation from the onset zone to other recording sites at seizure onset, as well as by changes in propagation between the onset zone and other recording sites during post-ictal intervals. This project was funded by NINDS R01 NS40596.