Amplitude of high frequency oscillations as a biomarker of the seizure onset zone

Abstract

Objective: Studies of high frequency oscillations (HFOs) in epilepsy have primarily tested the HFO rate as a biomarker of the seizure onset zone (SOZ), but the rate varies over time and is not robust for all individual subjects. As an alternative, we tested the performance of HFO amplitude as a potential SOZ biomarker using two automated detection algorithms.Method: HFOs were detected in intracranial electroencephalogram (iEEG) from 11 patients using a machine learning algorithm and a standard amplitude-based algorithm. For each detector, SOZ and non-SOZ channels were classified using the rate and amplitude of high frequency events, and performance was compared using receiver operating characteristic curves. Results:The amplitude of detected events was significantly higher in SOZ. Across subjects, amplitude more accurately classified SOZ/non-SOZ than rate (higher values of area under the ROC curve and sensitivity, and lower false positive rates). Moreover, amplitude was more consistent across segments of data, indicated by lower coefficient of variation. Conclusion:As an SOZ biomarker, HFO amplitude offers advantages over HFO rate: it exhibits higher classification accuracy, more consistency over time, and robustness to parameter changes.Significance: This biomarker has the potential to increase the generalizability of HFOs and facilitate clinical implementation as a tool for SOZ localization.

Refining epileptogenic high-frequency oscillations using deep learning: a reverse engineering approach

“…In particular, based on the analysis of time-frequency plots of eHFOs and non-eHFOs, eHFOs showed stronger signals throughout ripple and fast ripple bands at the onset of HFOs, which shares the similar observation in HFOs seen at the SOZ. 49 Meanwhile, eHFOs generally showed stronger signals in the low frequency region throughout the time window, which might represent an inhibitory slow-wave postsynaptic component, coupled with out-of-phase excitatory fast firing of HFOs. 50 Taking advantage of these observations, we designed perturbations in the input to probe whether these characteristics were actually the salient features that the model relied on to make a prediction.…”

Epileptogenic high-frequency oscillations present larger amplitude both in mesial temporal and neocortical regions

“…One possible approach is to examine HFO morphological features, such as amplitude, duration, and spectral frequency (Chen et al, 2021). When comparing HFO events detected in the epileptogenic vs. nonepileptogenic tissues, ripples showed larger amplitude, longer duration and lower frequency (von Ellenrieder et al, 2016, Charupanit et al, 2020, FR consistently showed larger amplitude but either shorter (Pail et al, 2017) or longer duration (von Ellenrieder et al, 2016) in the MTL, while neocortical HFO detected in the full spectral range showed either larger (Guragain et al, 2018) or lower amplitude (Alkawadri et al, 2014). In this study, rather than focusing on the statistical evaluation of the individual HFO morphological feature, we considered their multivariate information.…”

Refining epileptogenic high-frequency oscillations using deep learning: a reverse engineering approach

“…In particular, based on the analysis of time-frequency plots of eHFOs and non-eHFOs, eHFOs showed stronger signals throughout ripple and fast ripple bands at the onset of HFOs, which shares the similar observation in HFOs seen at the SOZ. 48 Meanwhile, eHFOs generally showed stronger signals in the low frequency region throughout the time window, which might represent an inhibitory slow-wave postsynaptic component, coupled with out-of-phase excitatory fast firing of HFOs. 49 Taking advantage of these observations, we designed perturbations in the input to probe whether these characteristics were actually the salient features that the model relied on to make a prediction.…”