Individualized epidemic spreading models predict epilepsy surgery outcomes: A pseudo-prospective study

Area of Science:

• Computational Neuroscience
• Epileptology
• Medical Imaging and Data Analysis

Background:

• Epilepsy surgery is a key treatment for drug-resistant epilepsy, yet many patients experience persistent seizures post-resection.
• Accurate presurgical planning is crucial for optimizing surgical outcomes and predicting patient-specific results.
• Current methods often struggle to predict postsurgical seizure control effectively.

Purpose of the Study:

• To develop and validate an individualized computational model, the Epidemic Spreading Seizure and Epilepsy Surgery (ESSES) framework, for predicting epilepsy surgery outcomes.
• To assess ESSES's ability to aid presurgical planning by identifying optimal resection strategies and estimating the likelihood of seizure freedom.
• To evaluate ESSES's performance in a pseudo-prospective, blinded setting without requiring invasive electroencephalography (iEEG) data.

Main Methods:

• Developed the ESSES framework combining epidemic spreading principles with patient-specific connectivity and epileptogeneity maps.
• Fitted ESSES parameters retrospectively using iEEG-recorded seizures from 15 patients to reproduce seizure dynamics.
• Validated ESSES in a pseudo-prospective study of 34 patients, assessing its predictive accuracy for surgical outcomes (seizure-free vs. non-seizure-free) and comparing model-based resections with actual surgical plans.

Main Results:

• ESSES accurately reproduced iEEG-recorded seizures, performing better for seizure-free (SF) than non-seizure-free (NSF) patients.
• The model predicted good outcomes with 80.8% accuracy for SF and 75% for NSF cases in the pseudo-prospective validation.
• Model-based optimal resections were smaller for SF patients, and actual surgical plans showed greater overlap and impact on seizure propagation for SF patients compared to NSF patients.

Conclusions:

• Individualized computational models like ESSES can significantly inform epilepsy surgery planning by suggesting optimal resections and predicting postsurgical outcomes.
• ESSES demonstrates clinical applicability by successfully predicting outcomes in a blinded, pseudo-prospective cohort without requiring iEEG data.
• The findings suggest that ESSES can help identify patients likely to benefit from surgery and guide more effective resection strategies, potentially improving seizure control.