Causal Brain Network Evaluates Surgical Outcomes in Patients with Drug-Resistant Epilepsy

Abstract

Network neuroscience has greatly facilitated epilepsy studies; meanwhile, drug-resistant epilepsy (DRE) is increasingly recognized as a brain network disorder. Unfortunately, surgical success rates in patients with DRE are still very limited, varying 30% [~] 70%. At present, there is almost no systematic exploration of intracranial electrophysiological brain network closely related to surgical outcomes, and it is not clear which brain network methodologies can effectively promote DRE precision medicine. In this retrospective comparative study, we included multicenter datasets, containing electrocorticogram (ECoG) data from 17 DRE patients with 55 seizures. Ictal ECoG within clinically-annotated epileptogenic zone (EZ) and non-epileptogenic zone (NEZ) were separately computed using six different algorithms to construct causal brain networks. All the brain network results were divided into two groups, successful and failed surgery. Statistical results based on the Mann-Whitney-U-test show that: causal connectivity of -frequency band (8 [~] 13 Hz) in EZ calculated by convergent cross mapping (CCM) gains the most significant differences between the surgical success and failure groups, with a P value of 7.85e-08 and Cohens d effect size of 0.77. CCM-defined EZ brain network can also distinguish the successful and failed surgeries considering clinical covariates (clinical centers, DRE types) with P < 0. 001. Based on the brain network features, machine learning models are established to predict the surgical outcomes. Among them, SVM classifier with Gaussian kernel function and Bayesian Optimization demonstrates the highest average accuracy of 84.48% through 5-fold cross validation, further indicating that the CCM-defined EZ brain network is a reliable biomarker for predicting DREs surgical outcomes.