Combining Spatial Sampling of Intracranial EEG and Ictal Epileptogenicity Improves Focality Prediction
Abstract number :
1.298
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2024
Submission ID :
1191
Source :
www.aesnet.org
Presentation date :
12/7/2024 12:00:00 AM
Published date :
Authors :
Presenting Author: Sarah Lavelle, MEng – University of Pennsylvania
Akash Pattnaik, BS – University of Pennsylvania
Ryan S Gallagher, MD – University of Pennsylvania
Alfredo Lucas, PhD – University of Pennsylvania
Daniel Zhou, MD – University of Pennsylvania
Joel Stein, MD, PhD – University of Pennsylvania
Kathryn Davis, MD – University of Pennsylvania
Erin Conrad, MD – University of Pennsylvania
Nishant Sinha, PhD – University of Pennsylvania
Brian Litt, MD – University of Pennsylvania
Rationale: For patients with refractory epilepsy, intracranial EEG (iEEG) during presurgical evaluation is crucial to determine intervention sites and assess eligibility for focal treatment. Current qualitative assessment methods vary across centers, highlighting the need for standardized quantitative approaches to accurately identify the focality of epileptic seizures and tailor appropriate treatments.
Methods: We analyzed data from 97 patients with drug-resistant epilepsy who underwent presurgical evaluation using iEEG. Among these, 65 patients had unifocal seizure onset, while 32 exhibited multifocal or diffuse seizures. We extracted features for each patients’ seizures using the epileptogenicity index (EI), which measures fast oscillation epileptogenicity on each channel (Bartolomei et al., 2008). To overcome varying spatial density of electrodes across patients and summarize the channel-level epileptogenicity, we computed the weighted standard distance (WSD) (Gallagher et al., 2023) of EI for each seizure. This novel combination, termed the EI-weighted-standard-distance (EI-WSD), was compared against the “5 Sense Score” (5SS)—a pre-implant estimate of focal seizure likelihood—and WSD of interictal abnormalities on the same cohort. We compared these features based on their area under receiver operating characteristic curve (AUC) from logistic regression classifiers to predict focality, with labels for focality obtained from clinicians' notes.
Results: The EI-WSD effectively distinguished between focal and non-focal patients (p< 0.001, Cohen’s d = 0.89) in the cohort of 97 patients. Electrode placement alone predicted network focality with a precision comparable to the 5SS (AUC = 0.64 vs. AUC = 0.65), indicating that electrode placement accurately reflected pre-implant information. Moreover, the EI-WSD outperformed both the 5SS and the spatial dispersion of interictal iEEG abnormalities in predicting network focality (AUC = 0.70). Combining the EI-WSD with WSDs of interictal delta and gamma bandpower abnormalities significantly enhanced prediction accuracy (AUC = 0.79).
Conclusions: A quantitative measure of iEEG seizure focality outperforms prediction of focality compared to pre-implant, noninvasive hypotheses. This study represents a positive step towards the establishment of standardized quantitative tools to guide epilepsy surgery, promising enhanced outcomes for patients through more precise treatment strategies.
Funding: The Mirowski Family Foundation, Jonathan and Bonnie Rothberg and R01-NS-125137-01.
Neurophysiology