Abstracts

Rationale: Until today, few studies describe electrical source localization in patients suffering from extra-temporal lobe epilepsy (ETLE), which is generally accepted to be more challenging to treat than temporal lobe epilepsy (TLE). In a group of ETLE patients exclusively, we determined the accuracy to localize the seizure onset zone (SOZ) and the irritative zone (IZ) by using an automated ictal and interictal Electrical Source Imaging (ESI), respectively.  Methods: We retrospectively included patients who underwent ETLE resective surgery. Epileptologists marked seizure onsets showing a clear EEG rhythmical activity. A spectrogram-based algorithm allowed automated selection of the parameters of analysis (Figure 1). Two ictal analyses were compared: (i) ESI power, where we considered as SOZ the source with the highest power, (ii) ESI + functional connectivity (FC), where we used the source with the highest out-degree, based on Granger causality FC analysis. Automated interictal spike-detection and ESI was also performed. The resection zone was used as reference standard for sublobar concordance. Accuracy was determined based on surgical outcome, at seizure and patient level, the latter involving a global clinical interpretation by the epileptologist.  Results: Ninety-four seizures from 24 patients were analyzed. At a seizure level, ESI power localized 38/94 seizures correctly, with a sensitivity of 36% and a specificity of 52%. ESI+FC significantly improved the performance, as 57/94 seizures were correctly localized (chi-square test, p = 0.0056), with a sensitivity of 52% and a specificity of 84%. Positive predictive value for ESI+FC was 90%. Results of ESI+FC were similar in patients with localizing, lateralizing or generalized ictal EEG (Kruskal-Wallis H test, p= 0.85) (Figure 2). Moreover, the size of the resection zone did not influence the performance of the method (Mann-Whitney U test, p= 0.53). At a patient level, 18/24 cases were studied, which included only focal resections and no hemispherectomies. ESI+FC showed a correct localization in 10/18 patients (55%), and interictal ESI in 11/18 (61%). In 5/18 patients, ictal ESI+FC could localize correctly the SOZ despite an erroneous localization of interictal ESI, whereas in 6/18 patients with inconclusive ictal analysis interictal ESI yielded correct results. Overall, only in two cases both ictal and interictal analyses failed to provide correct source localization.  Conclusions: We implemented an algorithm readily applicable on a 19-21 channel EEG that allows performing ESI and connectivity analysis on ictal EEG. FC significantly improves SOZ localization compared to ictal ESI solely in ETLE. Ictal and interictal ESI may have a complementary value. Our results encourage the use of both investigations in epilepsy pre-surgical workup.  Funding: No funding