Abstracts

Combined High-Frequency Oscillations and Inter-Ictal Epileptiform Sequences Predict Seizure Onset Zones

Abstract number : 1.033
Submission category : 1. Basic Mechanisms / 1C. Electrophysiology/High frequency oscillations
Year : 2023
Submission ID : 526
Source : www.aesnet.org
Presentation date : 12/2/2023 12:00:00 AM
Published date :

Authors :
Presenting Author: Christoph Kapeller, Dr. – g.tec medical engineering GmbH

Christoph Guger, PhD – g.tec medical engineering GmbH; Kyousuke Kamada, MD – Asahikawa Medical University; Christoph Kapeller, PhD – g.tec medical engineering GmbH; Fan Cao, MSc – g.tec medical engineering GmbH

Rationale:

As a biomarker for the epileptogenic zone, the inter-ictal epileptiform discharges (IED or spike) and high-frequency oscillations (HFOs) reflect the pathological strength of underlying tissues. Pathological HFOs (pHFO), a combination of spikes and HFOs, can help to define resection areas in patients with refractory epilepsy. Identifying the correct spike waveform is challenging, but by using spatiotemporal propagation of spikes across the epileptic cortex, we can better delineate the true spike [1]. Consequently, the epileptic zone can be predicted more accurately by pHFO than the spike rate itself [2]. In this work, the resting-state ECoG signals of three epileptic patients were analyzed to find spike sequence and HFOs, and with their combination, we tried to predict the seizure onset zone (SOZ) location of epileptic tissue.



Methods:

Three patients underwent intracranial EEG (iEEG) electrode implantation at the Megumino Hospital in Japan. The epilepsy monitoring unit (EMU) collected data when subjects were in their resting state of sleep at night. The SOZ was identified from ECoG data and verified with video EEG captured in EMU. An automatic event detector discovered HFOs and spikes events. Following individual spikes, we detected those spike sequences whose interictal spikes happened over several electrodes in close temporal proximity. The first spike was described as the sequence's leader. Spikes within 50 milliseconds of the leader or 15 milliseconds of the previous spike, were added to the spike sequence. After that, spike sequences were discarded if they contained less than five spikes. After detecting the spikes and HFOs separately, the time points were compared to see if they coincided, and if so, the event is referred to as a pHFO. Four biomarkers, namely, spike (SPK), spike using spike sequence analysis (SPK_SEQ), a combined spike with HFO (SPK_HFO), and a combined HFO with a spike in a spike sequence (SPK_SEQ_HFO), were calculated to find their relationship between with the SOZ in terms of the positive predictive value (PPV).



Results:

All biomarkers were detected using an automatic detection algorithm from 430 electrodes in three patients with 69.32 minutes of data in total. The SPK rate per electrode was 17.38, SPK_SEQ was 1.14, SPK_HFO was 3.52, and SPK_SEQ_HFO was 0.38 events per electrode on average. PPV values for SPK, SPK_SEQ, SPK_HFO, and SPK_SEQ_HFO were 0.72, 0.82, 0.87, and 0.91 on average, respectively.

 



Conclusions:

Clinically, the spike may be present in all brain regions and manifest in either SOZ or healthy areas. According to preliminary findings, pHFO, localized by combined spike sequences and HFOs, can predict SOZ more accurately than HFO or spike detection alone.

Funding:

The research was supported by the PowerMaps EU project (114618).



Basic Mechanisms