Mapping interictal discharges using intracranial EEG-fMRI to predict postsurgical outcomes
William Wilson1,2, Negar Tehrani1,2, Daniel J Pittman1,2
1Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary, Calgary, Alberta, T2N 4N1, Canada.
Brain : a Journal of Neurology
|May 9, 2024
Summary
Objective methods for analyzing EEG-fMRI data improve localization of seizure-generating tissue in epilepsy surgery. However, resecting the identified area alone does not guarantee seizure freedom, though failing to resect it predicts poor outcomes.
Area of Science:
- Neuroscience
- Medical Imaging
- Epileptology
Background:
- Accurate localization of seizure-onset zones is critical for successful epilepsy surgery.
- Interictal epileptiform discharges (IEDs) on EEG can guide functional MRI (fMRI) to identify potential seizure-generating regions.
- Existing methods for interpreting EEG-fMRI data vary in their ability to predict surgical outcomes.
Purpose of the Study:
- To evaluate subjective and objective methods for identifying clinically relevant EEG-fMRI activations for surgical planning.
- To compare the extent of resection of these identified regions with postoperative outcomes in drug-resistant focal epilepsy.
- To determine the predictive value of EEG-fMRI findings for surgical success.
Main Methods:
- Simultaneous intracranial EEG-fMRI was performed in 70 patients undergoing epilepsy monitoring.
- Clinically relevant and maximum EEG-fMRI activation clusters associated with IEDs were identified.
- Clusters were assigned confidence levels (low, medium, high) based on statistical significance.
- Spatial concordance with electrode contacts and resection areas was assessed.
- Postoperative outcomes were compared with the extent of resection of identified EEG-fMRI clusters.
Main Results:
- Both subjective and objective methods enhanced spatial concordance between EEG-fMRI clusters and seizure onset zones.
- Objective methods identifying medium and high confidence maps significantly associated resection of the maximum cluster's peak voxel with good, but not seizure-free, outcomes.
- Failure to resect medium and high confidence maximum clusters strongly predicted poor post-surgical outcomes (NPV=1.0, sensitivity=1.0).
Conclusions:
- Objective methods are crucial for improving the reliability of EEG-fMRI in localizing epileptogenic tissue.
- Resection of the peak voxel in high-confidence EEG-fMRI clusters is necessary but not sufficient for seizure freedom.
- Failure to resect these identified regions is a significant predictor of surgical failure, highlighting their importance.
