Abstracts

Rationale: Seizure detection and prediction aim to improve the quality of life in epilepsy patients. However, existing devices lack long-term seizure prediction capability. This study is an initial step in developing a novel endovascular seizure detection device. We aim to utilize SEEG data to determine the optimal venous location for an endovascular device.

Methods: We identified 6 venous sinus segments: anterior (SS-A), middle (SS-M) and posterior (SS-P) sagittal sinus, straight sinus, ipsilateral (ITS) (ipsilateral to the primary epileptogenic zone), and contralateral (CTS) transverse sinus (Figure 1). Distances between each sinus and SEEG contacts were measured. Data were categorized into seizure level (SL) (Focal-impaired-awareness (FIAS), Focal-aware (FAS) and Focal-to-bilateral-tonic-clonic (FBTCS) seizure) and epilepsy level (EL) (Temporal (TLE) and Extra-temporal (ETLE) lobe epilepsy). SEEG bipolar contacts were identified as “detect-a-seizure” if they were involved in the seizure. If those electrodes were within 5 cm from a sinus, the sinus will be defined as “detect-a-seizure”. We used multilevel mixed-effects regression and pairwise comparisons to determine the optimal sinus segment or the sinuses that can detect most seizures and the likelihood of being able to detect seizures at early stages (less than 5 s).


Results: This study involved 6,707 electrode contacts and 113 seizures (FIAS: 50.4%, FAS: 30.1%, FBTCS: 19.5%) in 40 SEEG patients (TLE: 50.0%, ETLE: 50.0%) .



At the SL, ITS showed the highest percentage of detectable seizures within 5 seconds for FAS (89.7%), FIAS (79.3%), and FBTCS (88.9%) (Figure 2). ITS outperformed other sinuses significantly for both FAS and FIAS except for the straight sinus. (p=0.65) in FIAS. In the FBTCS group, there was no significant difference between sinuses, except that ITS was better than SS-A (p=0.02). Regarding the likelihood of early seizure detection, ITS outperformed most sinuses for FAS, except SS-P (p=0.14) and SS-A (p= 0.16). In FIAS, SS-A showed the significantly lowest potential of detecting SZ. In FBTCS, there was no difference between sinuses .



At EL, ITS recorded the highest percentage (82.5% in TLE and 86.1% in ETLE) (Figure 2) of detectable seizures in the first 5s from seizure onset. For TLE, straight sinus (p=1.00) yielded a similar outcome in detecting seizures, without differences compared to ITS. SS-A showed the lowest potential to be able to detect seizure. In ETLE , SS-A, SS-M and SS-P showed capabilities in detecting seizures, without differences in pairwise comparison and likelihood testing. However, CTS showed the lowest potential to detect seizures in ETLE


Conclusions: Seizure type influences the number of detectable seizures near each venous sinus. As ITS consistently demonstrated better seizure detection capabilities across categories, identification of the ipsilateral side is crucial. Other sinuses, like the straight sinus (for FIAS and TLE) or sagittal sinus (for ETLE) should also be considered candidate targets.

Funding: University scholarship