Abstracts

Spatial Extent of Interictal Intracranial EEG Abnormalities Relates to the Focality of Epileptic Networks

Abstract number : 3.158
Submission category : 3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year : 2023
Submission ID : 801
Source : www.aesnet.org
Presentation date : 12/4/2023 12:00:00 AM
Published date :

Authors :
Presenting Author: Ryan Gallagher, BS – University of Pennsylvania

Nishant Sinha, PhD – University of Pennsylvania; Akash Pattnaik, BS – University of Pennsylvania; Alfredo Lucas, MS – Lucas, Alfredo; William Ojemann, BS – University of Pennsylvania; John Bernabei, MD, PhD – University of Pennsylvania; Joshua Laroque, MD, PhD – University of Pennsylvania; Elizabeth Sweeney, PhD – University of Pennsylvania; Isaac Chen, MD – University of Pennsylvania; Kathryn Davis, MD – University of Pennsylvania; Erin Conrad, MD – University of Pennsylvania; Brian Litt, MD – University of Pennsylvania

Rationale: Forty to sixty percent of patients with drug-resistant epilepsy relapse after epilepsy surgery. Assessing the distribution of each patient’s epileptic network is critical to surgical planning but lacks rigorous quantification. Here, we quantify the spatial extent of epilepsy biomarkers from intracranial EEG (IEEG), aiming to refine the presurgical hypothesis to predict the presence of a single focal seizure onset zone. We evaluate whether quantitative focality predictions of interictal EEG abnormalities relate to surgical decisions and patient outcomes.

Methods: We conducted a retrospective study on 101 patients undergoing IEEG for epilepsy surgery evaluation. Of these, 65 patients were clinically determined to have unifocal seizure onset on IEEG, and 36 were non-focal (i.e., multifocal, broad, bilateral). We used the 5-SENSE score to estimate the preimplantation likelihood of focal seizure onset, then quantified both the spatial coverage of implanted electrodes ("implant distances") and the spatial dispersion of interictal IEEG abnormalities ("abnormality distances") using an atlas of normative iEEG activity. We compared unifocal and non-focal patients using these features with univariate and multivariate analysis. We investigated if predictions of focality differentiated between patients who underwent surgery or device implantation and two year surgical outcomes in posthoc analysis.



Results: The 5-SENSE score, spatial coverage of IEEG, and spatial dispersion of IEEG abnormalities independently distinguish between focal and non-focal epileptic networks by univariate testing. Cross-validated Lasso regression combining the 5-SENSE score with IEEG abnormalities predicted the focality of epileptic networks with an AUC of 0.79, significantly better (DeLong’s test p< 0.05) than the 5-SENSE score alone (AUC=0.68) or 5-SENSE score with only the spatial coverage (AUC = 0.67). The probability of focal seizure onset zone predicted by this model could differentiate between the treatment strategies (surgery vs. device) with an AUC of 0.81 and seizure free from not seizure-free post-surgical seizure outcomes at 2 years with an AUC of 0.70.

Neurophysiology