T12. Automated EEG analysis of long-term EEG to localize the epileptogenic zone in extra-temporal lobe epilepsy

• Published in: Clinical neurophysiology 2018-05, Vol.129, p.e5-e6
• Main Authors: van Mierlo, Pieter, Baroumand, Amir Ghasemi, Keereman, Vincent, Strobbe, Gregor, Seeck, Margitta, Vulliémoz, Serge
• Format: Article
• Language: English
• ISSN: 1388-2457; 1872-8952
• DOI: 10.1016/j.clinph.2018.04.013

The localization of the epileptogenic focus in patients with extra temporal lobe epilepsy (ETLE) is often complex. Extra investigations such as ictal SPECT, interictal PET and MEG are frequently performed to obtain more information about where the epilepsy originates from. Nevertheless, these patients can have infrequent interictal epileptic discharges and seizures. Long-term video EEG monitoring allows recording interictal and ictal epochs over a longer period of time. In this study we use automated long-term EEG analysis to localize the epileptogenic focus in ETLE patients. We searched the database of Geneva University Hospital for patients that had ETLE surgery. We included patients that (i) had long-term EEG recording, (ii) had pre-operative MRI, (iii) post-operative follow-up of at least 1 years and (iv) no prior brain surgery. This resulted in the inclusion of 24 patients with ETLE. Of these patients, 15 had Engel class I and 9 had Engel class II, III or IV outcome. For all patients automated EEG analysis was performed by Epilog (Epilog NV, Ghent, Belgium). This included automated spike detection and subsequent EEG source imaging to localize the sources of the detected spikes. The spike clusters were scored by an expert electrophysiologist as genuine interictal discharges, artefact or physiological brain waves. Afterwards, the genuine spike clusters were localized at onset, half-rising time and peak of the spike in the brain using patient specific head model constructed from the patient’s MRI. The localization was compared to the resection at sub-lobar level to assess the sensitivity and specificity of the method to localize the epileptogenic zone. In 62% of patients (n = 15/24) genuine interictal spikes were detected by the software according to the scoring of the expert electrophysiologist. Of these 15 patients, 9 had Engel class I outcome and 6 had Engel class II, III or IV outcome. The localization of the spike cluster 1, i.e. the spike cluster with most detected single events, had a sensitivity of 22%, 44% and 44% and specificity of 100%, 83% and 83% to localize the epileptogenic focus at the onset, half-rising time and peak of the spikes, respectively. For spike cluster 2, i.e. the spike cluster with second most detected single events, the sensitivity to localize the epileptogenic focus was 25%, 25% and 0%; while the specificity was equal to 86%, 71% and 71% at onset, half-rising and peak of the spike, respectively. Combining cluster 1 and 2 re