Automated ictal EEG source imaging: A retrospective, blinded clinical validation study. (September 2022)

• Record Type: Journal Article
• Authors: Baroumand, Amir G.; Arbune, Anca A.; Strobbe, Gregor; Keereman, Vincent; Pinborg, Lars H.; Fabricius, Martin; Rubboli, Guido; Gøbel Madsen, Camilla; Jespersen, Bo; Brennum, Jannick; Mølby Henriksen, Otto; Mierlo, Pieter van; Beniczky, Sándor
• Is Part Of: Clinical neurophysiology. Volume 141(2022)
• Page Start: 119
• Page End: 125
• Publication Date: 2022-09
• Subjects: Automated -- EEG -- Epilepsy surgery -- Source analysis -- Source imaging -- Spectral analysis; Neurophysiology -- Periodicals; Electroencephalography -- Periodicals; Electromyography -- Periodicals; Neurology -- Periodicals; 612.8
• DOI: 10.1016/j.clinph.2021.03.040
• Languages: English
• ISSNs: 1388-2457

Abstract

Highlights: We have developed an automated pipeline for ictal EEG source imaging (ESI). We have analyzed ictal EEG signals from 50 operated patients with drug-resistant focal epilepsy. The automated ESI had an accuracy of 74% (95% CI: 59.66–85.37%). Abstract: Objective: EEG source imaging (ESI) is a validated tool in the multimodal workup of patients with drug resistant focal epilepsy. However, it requires special expertise and it is underutilized. To circumvent this, automated analysis pipelines have been developed and validated for the interictal discharges. In this study, we present the clinical validation of an automated ESI for ictal EEG signals. Methods: We have developed an automated analysis pipeline of ictal EEG activity, based on spectral analysis in source space, using an individual head model of six tissues. The analysis was done blinded to all other data. As reference standard, we used the concordance with the resected area and one-year postoperative outcome. Results: We analyzed 50 consecutive patients undergoing epilepsy surgery (34 temporal and 16 extra-temporal). Thirty patients (60%) became seizure-free. The accuracy of the automated ESI was 74% (95% confidence interval: 59.66–85.37%). Conclusions: Automated ictal ESI has a high accuracy for localizing the seizure onset zone. Significance: Automating the ESI of the ictal EEG signals will facilitate implementation of this tool in the presurgical evaluation.