Functional Connectivity Estimated Through Virtual Implantation Predicts Surgical Outcome in Children with Epilepsy
Abstract number :
1.109
Submission category :
2. Translational Research / 2C. Biomarkers
Year :
2021
Submission ID :
1825727
Source :
www.aesnet.org
Presentation date :
12/4/2021 12:00:00 PM
Published date :
Nov 22, 2021, 06:50 AM
Authors :
Ludovica Corona, MSc - University of Texas at Arlington, Arlington, TX, USA; Eleonora Tamilia, PhD - Division of Newborn Medicine, Department of Medicine - Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA; M Scott Perry, M.D - Jane and John Justin Neurosciences Center - Cook Children’s Health Care System, Fort Worth, TX, USA; Joseph R Madsen, M.D - Division of Epilepsy Surgery, Department of Neurosurgery - Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA; Steven Stufflebeam, M.D - Athinoula Martinos Center for Biomedical Imaging - Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; Phillip L Pearl, M.D - Division of Epilepsy and Clinical Neurophysiology, Department of Neurology - Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA; Christos Papadelis, PhD - Director of Research, Jane and John Justin Neurosciences Center, Cook Children’s Health Care System, Fort Worth, TX, USA
Rationale: Functional Connectivity (FC) is a useful biomarker of the epileptogenic zone (EZ), describing how different brain areas are functionally connected based on their synchronous temporal activity. With the hypothesis that increased FC is linked to the EZ, here we aim to estimate FC noninvasively with electric and magnetic source imaging (ESI and MSI) via “implantation” of virtual sensors (VSs), and to assess if removing hubs (i.e., electrodes with the highest FC values) results in seizure-free outcomes.
Methods: We retrospectively analyzed intracranial EEG (iEEG), high-density EEG (HD-EEG), and magnetoencephalography (MEG) data from 37 children with drug resistant epilepsy (DRE) who underwent surgery. We performed ESI and MSI on HD-EEG and MEG data respectively and build VSs at the iEEG electrode locations. We analyzed one minute of data with and without epileptiform spikes and computed two FC measures [i.e., Amplitude Envelope Correlation (AEC) and Phase Locking Value (PLV)] in different frequency bands: delta (1-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), beta (12-30 Hz), and low gamma (30-50 Hz). A brain network was generated from each FC matrix using the Minimum Spanning Tree. For each node, we computed four centrality metrics: betweenness, closeness, degree, and eigenvector. Each metric was normalized by its maximum value. We compared FC metrics inside vs. outside resection in good (n=22) and poor outcome (n=15) patients and then estimated their ability to predict outcome, by testing whether the removal of the network hubs leads to the patient’s good outcome (Fisher’s exact test).
Results: We found higher FC inside than outside resection (p< 0.05) in good outcome patients for the following metrics and frequency bands: (i) iEEG: AEC (theta, alpha, and beta) and PLV (alpha) on data with spikes, AEC (theta, alpha, and gamma) and PLV (alpha) on data without spikes; (ii) ESI: AEC (theta) on data with spikes, AEC (alpha, beta, and gamma) and PLV (gamma) on data without spikes; and (iii) MSI: AEC (beta) on data with and without spikes, and PLV in all frequency bands. We also found increased FC inside than outside resection in good outcome patients for all centrality metrics in all frequency bands (p< 0.05). No differences were observed for poor outcome patients. Two metrics showed differences (p< 0.05) inside vs. outside resection in good outcome patients, across all analyses (Fig. 1); one of them (i.e., AEC betweenness in beta) was also predictive (p< 0.05) at the patient-level for all modalities. PLV closeness in alpha was mapped at the patient’s level to show changes of FC values based on the surgical outcome (Fig. 2).
Translational Research