Using stereo-electroencephalography data to model the optimal intracranial venous sinus location for an endovascular seizure detection device: A feasibility study.

objectiveWe investigated the theoretical optimal venous location for an endovascular seizure detection device using stereo-EEG (SEEG) data.

methodsSix venous sinus segments from consecutive SEEG patients: anterior, middle, and posterior sagittal sinus (SS-A, SS-M, SS-P), straight sinus, and the ipsilateral/contralateral (relative to primary epileptogenic zone) transverse sinus (ITS, CTS). 25 mm was selected as optimal propagation distance, satisfying two criteria: (1) >80% of seizures propagated to the reference point and (2) >90% of patients had at least one electrode within this range. A sinus was classified as "detecting" a seizure if SEEG contacts within a 25 mm radius showed ictal activity at <5 s, <10 s, or any time from onset. We analyzed data at seizure level (focal-preserved-consciousness seizure (FPC), focal-impaired-consciousness seizure (FIC), and focal-to-bilateral-tonic-clonic (FBTCS)) and epilepsy level (temporal/extratemporal (TLE/ETLE)). Multilevel regression and pairwise comparisons were performed to assess detection rates across sinuses.

resultsWe analyzed 113 seizures (FPC: 30.1%, FIC: 50.4%, FBTCS: 19.5%) from 40 patients (TLE: 50%). At the optimal propagation distance, 86.11% of seizures propagated to the reference point. Additionally, 92.5% of seizures were detectable from at least one sinus location at any time point, with an average minimum sinus-to-contact distance of 9.38 mm. Seizure level: ITS had the highest seizure detection rate at any time point (FPC: 71.43%, FIC: 95.56%, FBTCS: 100%) and was superior or non-inferior to other sinuses for FPC and FIC. No significant differences were observed for FBTCS between sinuses. Epilepsy-level: ITS detected the most seizures at any time point (TLE: 93.75%, ETLE: 82.35%). ITS outperformed other sinuses for TLE, except the straight sinus. For ETLE, only SS-P was significantly less effective than SS-M and SS-A. SIGNIFICANCE: Based upon SEEG seizure propagation modeling, over 90% of seizures were detectable from venous sinus locations, supporting the feasibility of endovascular seizure detection devices. ITS demonstrated a high yield for seizure detection across seizure and epilepsy types. PLAIN LANGUAGE SUMMARY: This study aims to investigate whether seizures can be detected from venous locations and to determine the best placement for endovascular devices. Over 90% of seizures were detectable within 25 mm of the venous sinus, supporting the feasibility of this approach. The most effective venous sinus depended on the type of seizure and epilepsy. The transverse sinus on the same side as the seizure origin (ipsilateral transverse sinus, ITS) showed the highest overall potential for seizure detection. Sagittal sinus-middle (SS-M) was the best for detecting focal-preserved-consciousness seizures, while ITS and straight sinus were the best for detecting focal-impaired-consciousness seizures.