Abstracts

Rationale: We evaluated whether deep learning applied to whole brain pre-surgical structural connectomes could be used to predict post-operative seizure outcome more accurately than inference from clinical variables in patients with mesial temporal lobe epilepsy (TLE). Methods: 50 patients with unilateral TLE were classified either as having persistent disabling seizures (SZ) or becoming seizure-free (SZF) at least one year after epilepsy surgery. Their pre-surgical structural connectomes were reconstructed from whole brain diffusion tensor imaging. A deep network was trained based on connectome data to classify seizure outcome using 5-fold cross-validation. Results: Classification accuracy of our neural network trained showed PPV (seizure-freedom) of 88±7% and mean NPV (seizure-refractoriness) of 79±8%. On the contrary, a classification model based on clinical variables alone yielded less than 50% accuracy. The specific features that contributed to high accuracy classification of the neural network were located not only in the ipsilateral temporal and extra-temporal regions, but also in the contralateral hemisphere. Conclusions: Deep learning demonstrated to be a powerful statistical approach capable of isolating abnormal individualized patterns from complex datasets to provide a highly accurate prediction of seizure outcomes after surgery. Features involved in this predictive model were both ipsilateral and contralateral to the clinical foci and spanned across limbic and extra-limbic networks. Funding: None