Deep Learning-based Fully Automated Brain Segmentation for Assessment of Structural Alteration of Brain in Epileptic Children with SCN1A Mutation
Abstract number :
2.107
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2023
Submission ID :
556
Source :
www.aesnet.org
Presentation date :
12/3/2023 12:00:00 AM
Published date :
Authors :
Presenting Author: Hyun-jin Kim, MD – Asan Medical Center Children’s Hospital, Ulsan University College of Medicine
Min-Jee Kim, MD – Department of Pediatrics – Asan Medical Center Children’s Hospital, Ulsan University College of Medicine; Tae-Sung Ko, MD – Asan Medical Center Children’s Hospital, Ulsan University College of Medicine; Mi-Sun Yum, MD – Asan Medical Center Children’s Hospital, Ulsan University College of Medicine
Rationale:
SCN1A gene encodes the alpha 1 subunit of the voltage gated sodium channel and mutations in SCN1A gene are related to several epilepsy syndromes including genetic epilepsy with febrile seizures plus and severe myoclonic epilepsy of infancy. One previous study reported structural changes of the brain in patients with SCN1A mutation including decreased total, subcortical brain volume, gray and white matter volume and mean cortical thickness. The purpose of this study was to investigate the use of deep learning-based fully automated brain segmentation for assessment of structural alteration of the brain in epileptic children with SCN1A mutation.
Methods:
This retrospective study was conducted with 3D T1-weighted MR images of total 120 subjects (14 children with SCN1A mutation and epilepsy and 106 normal control subjects). Brain volumes (total, subcortical and regional) and cortical thickness measurements were obtained with a deep learning-based fully automated brain segmentation analysis software. All measured volumes from each subjects were adjusted and normalized using their intracranial volumes (ICVs) so that the covariance of the adjusted volumes in the normal subjects and their ICVs are minimized.
Results:
Total brain, gray matter, cortical and subcortical gray matter volumes were significantly smaller in children with epilepsy and SCN1A mutation, compared to that of the normal control group. Among subcortical regions, putamen and caudate volumes were significantly reduced compared to the normal control group. From a regional analysis, volumes of left pars triangularis in frontal lobe, right inferior parietal cortex, right precuneus cortex and right middle temporal gyrus were significantly reduced in children with epilepsy and SCN1A mutation than the normal control group.
Conclusions: Deep learning-based fully automated brain segmentation was able to demonstrate developmental brain changes in children with epilepsy and SCN1A mutation.
Funding: Funding information is not applicable
Neurophysiology