Abstracts

Rationale: In drug-resistant epilepsy, reliable identification of the epileptogenic zone (EZ) during the presurgical evaluation is of high importance. Most interictal, intracranial electroencephalography (iEEG) studies focus on static evaluation of EEG features, e.g. average rate of interictal epileptic discharges (IEDs), typically calculated in iEEG segments not longer than 30 minutes. However, EEG in general is known for its non-stationary character and recent studies on chronic-EEG suggest an analysis of prolonged segments (in terms of days, weeks, or even months). In this study, we hypothesise that besides well described static electrophysiological differences between EZ and non-EZ regions, valuable information lies in the dynamic variability of specific iEEG features extracted from prolonged recordings.

Methods: We analyzed iEEG recordings (with 5 kHz sampling rate) from 7 consecutive patients with drug-resistant, focal epilepsy, recorded at the Brno Epilepsy Center (St Anne’s Hospital) between May 2017 and May 2019. All patients had an excellent post-surgical outcome Engel1A, with > 2-year follow-up. EZ was defined as the seizure onset zone (SOZ) marked at the seizure onset by certified epileptologists. IEDs (spikes) were detected by a validated detector and relative entropy was calculated as a nearest-neighbour bivariate measure of signal spectral complexity in a wide frequency range of unfiltered raw recordings. The median and variance of spike rates and relative entropy values were calculated for SOZ and non-SOZ contacts separately. The differences between SOZ and non-SOZ were tested by Mann-Whitney rank test. The p-values were corrected by the Bonferroni correction by factor 8 (2 regions, 4 tests).

Results: The average±std duration of 165,3±59,6 hours of continual interictal recordings was analyzed in 20 SOZ and 1002 non-SOZ iEEG contacts. Increased median spike rates were detected in the SOZ, compared to non-SOZ (p=0.012e-3), as well as increased median relative entropy (p=0.072e-3). Analysis of temporal dynamics of these features showed increased spike rate variance (p=0.01e-3) and relative entropy variance (p=0.000095e-3) in the SOZ, compared to non-SOZ regions.

Conclusions: The analysis of approximately one week long iEEG showed increased median spike rates and relative entropy values in the SOZ. Furthermore, and more importantly, the analysis of variance of these features showed that non-SOZ regions are significantly more stable than SOZ throughout the time of recording. This finding might contribute to the process of SOZ identification during presurgical planning and to our knowledge about the seizure onset regions in general.

Funding: Please list any funding that was received in support of this abstract.: This abstract was funded by the Ministry of Education, Youth and Sport of the Czech Republic, project INTER-EXCELLENCE - LTAUSA18056.