Abstracts

Rationale: Temporal lobe epilepsy (TLE) comprises most epilepsy surgical patients. Current literature characterizes TLE as a network disorder; network modeling with resting state fMRI reveals “anti-correlated” networks firing in opposition to the positively-correlated seizure network. The morphology of these networks predicts surgical outcomes, in that increased surgical disconnection of the seizure network and decreased disconnection of the anti-correlated network are associated with higher rates of seizure freedom and lower rates of post-operative neurocognitive deficits. Seizure semiology is also used to characterize seizure networks in TLE, but it remains a relatively untapped resource in predicting surgical outcomes. Here, we assess the relationship between semiology and positively- and anti-correlated networks using rsfMRI.

Methods: We prospectively selected 24 adult TLE surgical patients. We used each patient’s rsfMRI and scalp EEG to model their epileptogenic zones (EZ) and pre-operative seizure networks (both positively- and anti-correlated networks). Next, we characterized the patients’ semiologies according to 9 TLE signs. Principal component analysis of potential hypotheses generated outputs suggesting a potential relationship between the presence of head version (both ipsilateral and contralateral) and certain characteristics of the pre-operative anti-correlated seizure network. We examined that relationship with a Spearman correlation analysis p< 0.05.