Detection of high-frequency oscillations by hybrid depth electrodes in standard clinical intracranial EEG recordings

Abstract

High-frequency oscillations (HFOs) have been proposed as a novel marker for epileptogenic tissue, spurring tremendous research interest into the characterization of these transient events. A wealth of continuously recorded intracranial electroencephalographic (iEEG) data is currently available from patients undergoing invasive monitoring for the surgical treatment of epilepsy. In contrast to data recorded on research-customized recording systems, data from clinical acquisition systems remain an underutilized resource for HFO detection in most centers. The effective and reliable use of this clinically obtained data would be an important advance in the ongoing study of HFOs and their relationship to ictogenesis. The diagnostic utility of HFOs ultimately will be limited by the ability of clinicians to detect these brief, sporadic, and low amplitude events in an electrically noisy clinical environment. Indeed, one of the most significant factors limiting the use of such clinical recordings for research purposes is their low signal to noise ratio, especially in the higher frequency bands. In order to investigate the presence of HFOs in clinical data, we first obtained continuous intracranial recordings in a typical clinical environment using a commercially available, commonly utilized data acquisition system and "off the shelf" hybrid macro-/micro-depth electrodes. These data were then inspected for the presence of HFOs using semi-automated methods and expert manual review. With targeted removal of noise frequency content, HFOs were detected on both macro- and micro-contacts, and preferentially localized to seizure onset zones. HFOs detected by the offline, semi-automated method were also validated in the clinical viewer, demonstrating that (1) this clinical system allows for the visualization of HFOs and (2) with effective signal processing, clinical recordings can yield valuable information for offline analysis. © 2014 Kondylis, Wozny, Lipski, Popescu, DeStefino, Esmaeili, Raghu, Bagic and Richardson.