Spatial Patterns of Spread and Morphology of Seizures and Interictal Spikes Identify Patients with Mesial Temporal Lobe Epilepsy
Abstract number :
2.194
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2024
Submission ID :
929
Source :
www.aesnet.org
Presentation date :
12/8/2024 12:00:00 AM
Published date :
Authors :
Presenting Author: Carlos Aguila, BS – University of Pennsylvania
Alfredo Lucas, PhD – University of Pennsylvania
Sarah Lavelle, MEng – University of Pennsylvania
Akash Pattnaik, BS – University of Pennsylvania
Devin Ma, BS – University of Pennsylvania
William Ojemann, BS – University of Pennsylvania
Mariam Josyula, BA – University of Pennsylvania
Joshua LaRocque, MD, PhD – University of Pennsylvania
Ezequiel Gleichgerrcht, MD, PhD – Emory University
Colin Ellis, MD – University of Pennsylvania
Kathryn Davis, MD – University of Pennsylvania
Brian Litt, MD – University of Pennsylvania
Erin Conrad, MD – University of Pennsylvania
Rationale: Mesial temporal lobe epilepsy (mTLE) is one of the most common localizations of drug-resistant epilepsy in adults (Milligan, 2021). To confirm localization and to determine the candidacy for a focal ablation or resection, these patients often undergo intracranial EEG (iEEG) monitoring. Quantitative approaches have been proposed to characterize the timing and morphology of both seizures and interictal spikes. Here we combine these approaches to predict a localization of mesial temporal lobe epilepsy.
Methods: We conducted a retrospective cohort study of patients with drug-resistant epilepsy who underwent intracranial EEG monitoring at University of Pennsylvania (HUP) and Medical University of South Carolina (MUSC). An automatic, previously validated spike detection algorithm identified interictal spikes (Brown et al., 2007; Conrad et al., 2020), from which we calculated the timing of spikes in a sequence as well as multiple morphological features. Additionally, we computed the epileptogenicity index (EI) (Bartolomei et al., 2008) from seizure clips. We computed Pearson correlations across electrode contacts traversing the temporal lobe in a medial to lateral direction to represent spatial patterns (e,g, contact LA1 was compared with LA2, etc.). We compared spatial patterns across morphological features, spike rates, and EI between patients with mesial temporal, temporal neocortical, and other localizations. We then used all features to train, validate, and test Logistic regression (LR) classifiers using a nested Leave-One-Out-Cross-Validation approach to predict mTLE patients.
Results: We studied 77 patients (51 HUP, 26 MUSC). Spike rate (one-way ANOVA: p < 0.001) and spike morphology (decay amplitude and sharpness were significant after Benjamini-Hochberg correction) differed across epilepsy localizations. EI and timing did not differ significantly across localizations (p > 0.05). A logistic regression model combining all features predicted a diagnosis of mTLE in unseen HUP patients with an AUC of 0.78.
Conclusions: Our findings indicate that the spatial patterns of spread and the morphology of interictal spikes and seizures vary depending on the epilepsy localization and can predict a localization of mTLE. Univariate analyses suggest that interictal data may be more important than data from the seizures themselves at revealing mTLE localization. These results can enhance epileptic diagnosis and assist clinicians in making informed decisions regarding interventions targeting the mesial temporal lobe.
Funding:
Carlos Aguila received funding from the NSF Graduate Research Fellowships Program (GRFP), Brian Litt's R01-NS-125137 & DP1-NS-122038, and Erin Conrad's National Institute of Neurological Disorders and Stroke (NINDS K23 NS121401-01A1) and the Burroughs Wellcome Fund.
Neurophysiology