Transforming Clinical Practice in Epilepsy: A Novel Methodology for Enhanced Detection and Localization of Epileptiform Events
Abstract number :
1.241
Submission category :
3. Neurophysiology / 3D. MEG
Year :
2025
Submission ID :
990
Source :
www.aesnet.org
Presentation date :
12/6/2025 12:00:00 AM
Published date :
Authors :
Presenting Author: Velmurugan Jayabal, MBBS, PhD – University of California San Francisco
Heidi Kirsch, MD – University of California San Francisco
anne Findlay, BS – University of California San Francisco
Rachel Lentner, R. EEG T. – University of California San Francisco
Mary Mantle, EEG – University of California San Francisco
Robert Knowlton, MD, MSPH – University of California, San Francisco
Srikantan Nagarajan, PhD – University of California San Francisco
Rationale: The review of magnetoencephalography (MEG) data in epilepsy presents significant challenges, particularly in detecting and localizing subtle epileptiform events originating from deep structures. Complex patterns such as polyspikes, paroxysmal fast activities, and generalized spike-wave discharges are often obscured by interference from vagal nerve stimulators and deep brain stimulators. Traditional sensor-level MEG analysis frequently oversimplifies these intricate dynamics by relying solely on dipole localization, which presumes a single focal source at each marker.
Methods: We pioneered a hybrid methodology that integrates spatial filtering with source-independent component analysis to enhance the detection of epileptiform events within the whole brain source space, rather than limiting the analysis to sensor-level data. This innovative approach was prospectively applied to a cohort of 40 patients referred to our MEG center, categorized by clinical neurophysiologists into three distinct classes:
• Class I: Detectable but difficult-to-localize events (e.g., polyspikes, beta bursts).
• Class II: Detectable but ambiguously localized events (e.g., multifocal or bilateral sources).
• Class III: Non-identifiable events owing to low signal-to-noise ratio (SNR).
Presumed epileptogenic zones (EZs) were determined using a combination of pre-surgical EEG, video telemetry, and neuroimaging, with a subset of patients (n=5) undergoing intracranial EEG (iEEG) for validation.
Results: Our hybrid approach significantly improved detection and localization accuracy, particularly in Class II (n=15; 38%) and Class III (n=16; 40%) scenarios, with Class I contributing (n=4; 10%). Remarkably, 88% of patients (n=35/40) demonstrated enhanced detection and localization, exhibiting strong concordance with presumed EZs (Cohen’s k = 0.79). For 5 out of 16 patients, this method directed the placement of iEEG electrodes, which would have otherwise posed challenges in spatial sampling. Cortical stimulation of these electrodes successfully elicited aura and seizures, thereby facilitating surgical decision-making.
Conclusions: The integration of our innovative approach into routine clinical practice marks a transformative advancement in standard clinical MEG reporting, minimizes false negatives in MEG interpretations as well as significantly enhancing detection and localization of complex epileptiform events. This improvement reduces the necessity for phase II evaluations and optimizes therapeutic strategies, especially in patients needing guidance for neuromodulatory or surgical interventions. Our methodology promises to be seamlessly incorporated into clinical workflows, offering superior diagnostic utility with improved localization accuracy for Class II and III MEG cases.
Funding: NA
Neurophysiology