Non-Invasive MEG Network Biomarkers to Guide Hippocampus-Sparing Surgery in Temporal Lobe Epilepsy
Abstract number :
2.19
Submission category :
3. Neurophysiology / 3D. MEG
Year :
2025
Submission ID :
868
Source :
www.aesnet.org
Presentation date :
12/7/2025 12:00:00 AM
Published date :
Authors :
Presenting Author: Guhan Seshadri, PhD – Cleveland Clinic
Hiroatsu Murakami, MD, PhD – Cleveland Clinic
Richard Burgess, MD, PhD – Cleveland Clinic
Andreas Alexopoulos, MD, MPH – Cleveland Clinic
Balu Krishnan, PhD – Cleveland Clinic
Rationale: Temporal lobe epilepsy (TLE) is the most prevalent form of drug-resistant focal epilepsy. While anteromesial temporal lobectomy—which involves resecting the hippocampus, amygdala, and adjacent structures—remains the most effective treatment for seizure control (Wiebe et al., 2001), it often results in cognitive impairments due to hippocampal removal (Helmstaedter et al., 2011). As epilepsy research increasingly shifts toward network-based approaches, there is a growing interest in identifying functional connectivity biomarkers that can support hippocampus-sparing surgical strategies without compromising seizure outcomes. In this study, we examine hippocampal FC patterns using presurgical resting-state magnetoencephalography (MEG) in patients who underwent either hippocampal resection or sparing procedures.
Methods: Resting-state MEG was collected from 25 TLE patients (19 with hippocampal resections) who underwent presurgical MEG, stereo-EEG, and successful resective surgery. MEG data were denoised using temporally extended signal space separation (tSSS) to mitigate artifacts from noise and head motion (Taulu & Hari, 2009). Source reconstruction was performed via an LCMV beamformer, extracting time series from 162 regions defined by the Virtual Epileptic Patient (VEP) atlas. Functional connectivity (FC) was estimated using Phase Locking Value (PLV) in the gamma (30–60 Hz) and high-gamma (60–90 Hz) bands. A network of interest was defined by the top 10% strongest contralateral hippocampal connections across patients, with ipsilateral homologues used for comparison. To assess regional influence on network dynamics, control centrality (Ci) was computed for each region in the network of interest (Khambhati et al., 2016).
Results: Significant differences in Ci were observed within the mesial temporal network of patients who underwent hippocampal resection. In the gamma band, the ipsilateral anterior hippocampus (p=0.017), amygdala (p=0.029), and entorhinal cortex (p=0.01) showed higher and positive Ci compared to their contralateral homologues, with similar effects observed in the high-gamma band—most notably in the anterior hippocampus (p=0.012) and entorhinal cortex (p=0.0002). In contrast, the hippocampal-sparing group showed elevated Ci only in the anterior insula within the high-gamma band (p=0.026). These findings suggest that mesial temporal regions exert a stronger desynchronizing influence in resected patients, emphasizing their central role in seizure-generating networks.
Conclusions: Our findings demonstrate that control centrality derived from non-invasive presurgical MEG can differentiate hippocampal-sparing from hippocampal-resected TLE patients, reflecting distinct mesial temporal network dynamics. These results highlight the potential of MEG-based network metrics to support surgical planning and inform the selection of candidates for hippocampus-sparing procedures. Further validation in larger cohorts with clinical outcome correlation is required.
Funding: NA
Neurophysiology