Page 139 - MEGIN Book Of Abstracts - 2023

significantly correlating SEEG contact close to zero Medicine, Philadelphia, Pennsylvania, USA
lag after correcting for multiple comparisons. All the
patients except one had at least one epileptic compo- ABSTRACT Epilepsy is a highly heterogeneous neu-
nent that was highly correlated (Spearman rho>0.3) rological disorder with variable etiology, manifesta-
with that of SEEG traces. MEG-ICs correlated well with tion, and response to treatment. It is imperative that
SEEG traces. The strength of correlation coefficients did new models of epileptiform brain activity account for
not depend on the depth of the SEEG contacts or the this variability, to identify individual needs and allow
clinical outcome of the patient. A significant proportion clinicians to curate personalized care. Here, we use a
of the MEG-ICs (n = 83/131) were localized in proximity hidden Markov model (HMM) to create a unique statis-
with their maximally correlating SEEG, within a mean tical model of interictal brain activity for 10 pediatric
distance of 20±12.18mm. Our research is the first to patients. We use magnetoencephalography (MEG) data
validate the MEG-retrieved beamformer IC sources acquired as part of standard clinical care for patients
against SEEG-derived ground truth in a simultaneous at the Children's Hospital of Philadelphia. These data
MEG-SEEG framework. Observations from the present are routinely analyzed using excess kurtosis map-
study suggest that non-invasive MEG source compo- ping (EKM); however, as cases become more complex
nents may potentially provide additional information, (extreme multifocal and/or polymorphic activity), they
comparable to SEEG in a number of instances. become harder to interpret with EKM. We assessed the
performance of the HMM against EKM for three patient
Keywords: Epileptogenic zone, MEG and intracranial EEG, groups, with increasingly complicated presentation.
Simultaneous MEG and SEEG, Source ICA, Virtual sensors The difference in localization of epileptogenic foci for
the two methods was 7 ± 2 mm (mean ± SD over all
NeuroImage (2022), Vol. 264 (36270623) (1 citation) 10 patients); and 94% ± 13% of EKM temporal markers
were matched by an HMM state visit. The HMM local-
izes epileptogenic areas (in agreement with EKM) and
Mapping Interictal activity in epilepsy using a provides additional information about the relationship
hidden Markov model: A magnetoencephalography between those areas. A key advantage over current
study (2023) methods is that the HMM is a data-driven model, so the
output is tuned to each individual. Finally, the model
Seedat, Zelekha A; Rier, Lukas; Gascoyne, Lauren E; output is intuitive, allowing a user (clinician) to review
Cook, Harry; Woolrich, Mark W; Quinn, Andrew J; the result and manually select the HMM epileptiform
Roberts, Timothy P L; Furlong, Paul L; Armstrong, Caren; state, offering multiple advantages over previous meth-
St Pier, Kelly; Mullinger, Karen J; Marsh, Eric D; Brookes, ods and allowing for broader implementation of MEG
Matthew J; Gaetz, William epileptiform analysis in surgical decision-making for
patients with intractable epilepsy.
Young Epilepsy, St Pier's Lane, Lingfield, RH7 6PW, UK; Sir Peter
Mansfield Imaging Centre, School of Physics and Astronomy, Keywords: epilepsy, hidden Markov model, interictal activ-
University of Nottingham, Nottingham, UK; Oxford Centre for ity, magnetoencephalography
Human Brain Activity, University Department of Psychiatry,
Warneford Hospital, Oxford, UK; Department of Radiology, Human brain mapping (2023), Vol. 44, No. 1 (36259549) (1
Children's Hospital of Philadelphia, Philadelphia, Pennsylva- citation)
nia, USA; Aston Brain Centre, Aston University, Birmingham,
UK; Pediatric Epilepsy Program, Division of Child Neurology,
CHOP, Philadelphia, Pennsylvania, USA; Centre for Human
Brain Health, School of Psychology, University of Birming-
ham, Birmingham, UK; Departments of Neurology and
Paediatrics, University of Pennsylvania Perelman School of

 ontents  Index 118
C