Brain sodium MRI-derived priors support the estimation of epileptogenic zones using personalized model-based methods in Epilepsy

Epilepsy patients with drug-resistant epilepsy are eligible for surgery aiming to remove the regions involved in the production of seizure activities (the so-called epileptogenic zone network (EZN). Thus the accurate estimation of the EZN is crucial. A data-driven, personalized virtual brain models derived from patient-specific anatomical and functional data are used in Virtual Epileptic Patient (VEP) to estimate the EZN via optimization methods from Bayesian inference. The Bayesian inference approach used in previous VEP integrates priors, based on the features of stereotactic-electroencephalography (SEEG) seizures’ recordings. Here, we propose new priors, based on quantitative23Na-MRI. The23Na-MRI data were acquired at 7T and provided several features characterizing the sodium signal decay. The hypothesis is that the sodium features are biomarkers of neuronal excitability related to the EZN and will add additional information to VEP estimation. In this paper, we first proposed the mapping from23Na-MRI features to predict the EZN via a machine learning approach. Then, exploiting these predictions as priors in the VEP pipeline, we demonstrated that23Na-MRI prior based VEP estimation of the EZN improved the results in terms of balanced accuracy and as good as SEEG priors in terms of the weighted harmonic mean of the precision and recall.AUTHOR SUMMARYFor the first time quantitative23Na-MRI were used as prior information to improve estimation of EZN using the model-based method of VEP pipeline. The priors were based on logistic regression predictions of the EZN, using23Na-MRI features as predictors. The23Na-MRI priors inferred EZNs significantly closer to the clinical hypotheses - in terms of balanced accuracy - than the previously used priors or the no-prior condition.