Automated Detector of High Frequency Oscillations in Epilepsy Based on Maximum Distributed Peak Points

Abstract

High frequency oscillations (HFOs) are considered as biomarker for epileptogenicity. Reliable automation of HFOs detection is necessary for rapid and objective analysis, and is determined by accurate computation of the baseline. Although most existing automated detectors measure baseline accurately in channels with rare HFOs, they lose accuracy in channels with frequent HFOs. Here, we proposed a novel algorithm using the maximum distributed peak points method to improve baseline determination accuracy in channels with wide HFOs activity ranges and calculate a dynamic baseline. Interictal ripples (80-200[Formula: see text]Hz), fast ripples (FRs, 200-500[Formula: see text]Hz) and baselines in intracerebral EEGs from seven patients with intractable epilepsy were identified by experienced reviewers and by our computer-automated program, and the results were compared. We also compared the performance of our detector to four well-known detectors integrated in RIPPLELAB. The sensitivity and specificity of our detector were, respectively, 71% and 75% for ripples and 66% and 84% for FRs. Spearman's rank correlation coefficient comparing automated and manual detection was [Formula: see text] for ripples and [Formula: see text] for FRs ([Formula: see text]). In comparison to other detectors, our detector had a relatively higher sensitivity and specificity. In conclusion, our automated detector is able to accurately calculate a dynamic iEEG baseline in different HFO activity channels using the maximum distributed peak points method, resulting in higher sensitivity and specificity than other available HFO detectors.

Optimizing automated detection of high frequency oscillations using visual markings does not improve SOZ localization

“…However, the use of such standard configurations leads to suboptimal performance when used for a different data set or frequency range. For example, when previously published algorithms are implemented without optimizing the parameters, they exhibit lower detection accuracy (relative to visually detected events) than originally reported (Gardner et al 2007;Dümpelmann et al 2012;Cimbálník et al 2018;Ren et al 2018). Consistent with this, several studies have demonstrated that HFO detection accuracy relative to visually marked events can be improved by optimizing the algorithm parameters (Zelmann et al 2012;Chaibi et al 2013;Charupanit and Lopour 2017).…”

Determining the Quantitative Threshold of High-Frequency Oscillation Distribution to Delineate the Epileptogenic Zone by Automated Detection

“…In order to detect HFOs, we selected five minute segments during slow sleep period in which the delta band measured higher than 25% of all delta bands in a 30 second epoch. We also refer to the results of electrooculography and chin electromyography in determining the slow wave sleep period ( 10 ). All segments were selected from interictal periods, separated at least 2 h from seizures, and were transformed to a bipolar montage made of adjacent contacts.…”

“…Over the past two decades, numerous studies have shown that the removal of areas showing high rates of high frequency oscillations (HFOs) is associated with a good-surgical outcome ( 4 – 8 ). Therefore, HFOs are considered a promising biomarker of the seizure onset zone (SOZ) or epileptogenic zone ( 4 , 9 , 10 ). However, none of these studies have applied the quantitative threshold of HFO distribution to delineate the EZ.…”

“…HFOs are classified into ripples (80–200 Hz) and fast ripples (FRs, 200–500 Hz). Although visual analysis remains the gold standard for HFO analysis, this procedure is time consuming and subjective ( 4 , 10 ). Therefore, developing an automated detector is critical ( 4 , 11 ).…”

“…When the channel contains massive HFOs or high-frequency activities, the calculated baseline level will deviate significantly from the true baseline level, resulting in a significant drop in the accuracy of an automated detection. To solve this problem, Ren et al proposed an algorithm that calculates the baseline by maximum distributed peak points, which worked well ( 10 ).…”