Abstracts

Rationale: Three-dimensional scalp EEG source imaging (3D-ESI) combines imaging modalities with EEG source analysis to localize seizure onset zones. It is a non-invasive, inexpensive tool that adds no additional burden to patients. We examined the concordance of 3D-ESI localization results produced by CURRY with regions of resected focal cortical dysplasia (FCD). 42% of epilepsy surgery patients have FCD; however, 23-73% of those patients are MRI-negative and need further studies to identify seizure foci. Few studies have explored CURRY in pediatric epilepsy. This study evaluates the agreement of CURRY localization with final surgical resection in FCD patients.

Methods: 16 patients (11 male, mean age at surgery = 9y, range 0-21y) who underwent surgery and had pathologically-proven FCD were included. We conducted retrospective blinded 3D source localization through CURRY V.8.0. For each patient, a 3D head model was created from pre-operative MRI and two seizures from last preoperative EEG. We compared CURRY results and surgical resections at the hemisphere and lobe levels to assess lateralization and localization accuracy. We used a novel analysis quantifying the percent of surgical resection that fell into CURRY current density maps. We report overlap as binary (“yes/no”) regardless of what percent of the resection is in the current density. We used chi-square analysis to test if the likelihood of seizure freedom at 6 months differed if CURRY results were concordant versus discordant.

Results: CURRY localized to surgical resection location with 100% accuracy (hemispheric) 75% accuracy (lobar level) confirmed through visual analysis). 81% of patients had some overlap between resection and current density map in either seizure; 56% had overlap across both seizures. 19% had no overlap between resection and CURRY. Percent of resection within the current density ranged from 0-100%. 13 patients had abnormal MRIs and 2 patients had FCDs that overlapped across two lobes, but their resections involved only one. There was no increased likelihood of seizure freedom at six months with concordant or discordant results.

Conclusions: CURRY accuracy at hemispheric and lobar levels aligns with traditional source localization methods and overlaps with surgical site. The relationship between CURRY signal extent, FCD, and resection require further study as we postulate CURRY may identify abnormal tissue beyond the resolution of current MRI. Source localization is fully concordant for hemisphere; however, this declines at the lobar level, and more specifically the surgical resection. Agreement was achieved in a majority of patients. A limitation of 3D-ESI localization is determining whether one is identifying seizure origin or propagation. This may explain discrepancies in our results. Future work should focus on specific metrics such as density map probability, percentage overlap, and CURRY-positive residual tissue; all of which will help to expand CURRY’s role in the presurgical evaluation.

Funding: Please list any funding that was received in support of this abstract.: No funding.