Abstracts

Rationale: Stereo-EEG (SEEG) has become the preferred methodology for intracranial EEG monitoring at epilepsy centers around the USA. Visual analysis of SEEG can be technically challenging, particularly to inexperienced users, and lead to subjective interpretations. While clinically meaningful interictal and ictal patterns can be visualized, translating these findings to a 3D representation, which may offer superior spatial characterization, requires quantitative techniques such as source imaging, seizure fingerprint and epileptogenic indices. In this study, we aimed to better characterize the value of such techniques in our institution in hopes of identifying specific cases or techniques that may have widespread application at other epilepsy centers and patient populations.


Methods: Brainstorm is an open-source application designed to analyze various brain recording modalities, including SEEG. The Brainstorm SEEG pipeline involved searching for seizure fingerprints on the back-projected time-series on the source level to identify the epileptogenic zone on the cortex. Source modeling was computed using sLORETA Min-Norm Imaging using a forward head model which was computed using boundary element modeling. Afterward, the Desikan-Killiany atlas was used to define multiple regions of interest (ROI), and the first principal component for each ROI was subject to Morlet wavelet transform to compute a spectrogram. All the spectrograms were screened for the presence of seizure fingerprint (Grinenko et at, 2018).

Results: At the Texas Comprehensive Epilepsy Program (TCEP), we have integrated Brainstorm into our SEEG invasive pipeline since May 2023. During this period, a total of 23 SEEG patients were evaluated. In select cases, Brainstorm was utilized. Commonly utilized techniques include performance of sensor level seizure fingerprint analysis and sLORETA source modeling of interictal spikes. Less commonly, source modeling of ictal rhythms were performed and in a case of a large malformation of cortical development, comparison of seizure fingerprint findings at different cortical parcellations was helpful in the ultimate resective strategy.

Conclusions: Advanced source and sensor level processing of SEEG data can significantly enhance the interpretation of intracranial EEG data, particularly in identifying the ictal onset zone in large cortical malformations, delineating the extent of tissue activation within dysplastic areas, and analyzing seizures with rapid propagation. Interictal spike analysis on a source level and ictal sensor or source level analysis can provide additional value beyond traditional visual analysis and were commonly utilized techniques. However, in cases where the ictal pattern was not as straightforward, source modeling of ictal spike rhythms or seizure fingerprint analysis on different cortical parcellations can provide additional information to help localize the epileptogenic zone. Further analysis on an institutional and multi-institutional level is needed to further understand the optimal utilization of such techniques.

Funding: None