Radiomics-based MRI Analysis Can Potentially Differentiate Patients with Juvenile Myoclonic Epilepsy from Healthy Controls
Abstract number :
2.183
Submission category :
5. Neuro Imaging / 5A. Structural Imaging
Year :
2022
Submission ID :
2204153
Source :
www.aesnet.org
Presentation date :
12/4/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:23 AM
Authors :
Hye Jeong Lee, MD – Severance Hospital, Yonsei University College of Medicine; Kyoung Heo, MD, PhD – Department of Neurology – Severance Hospital, Yonsei University College of Medicine; Kyung Min Kim, MD – Gangnam Severance Hospital, Yonsei University College of Medicine; Won-Joo Kim, MD, PhD – Gangnam Severance Hospital, Yonsei University College of Medicine
Rationale: This study aimed to build and validate radiomics prediction models that could discern patients with juvenile myoclonic epilepsy (JME) from healthy controls (HCs).
Methods: A total of 129 subjects (97 JME patients and 32 HCs) were assigned to a training (n=90) or a test set (n=39) group. Radiomics features were extracted from 20 regions of interest from T1-weighted magnetic resonance images (MRI). Several machine learning models were trained with the patients’ radiomics features during training sets, and they were validated with test sets.
Results: The seven tested radiomics models—light gradient boosting machine, support vector classifier, random forest, logistic regression, extreme gradient boosting, gradient boosting machine, and decision tree—showed an area under the curve of 0.82, 0.81, 0.78, 0.78, 0.77, 0.76, and 0.67, respectively. The best-performing model, the light gradient boosting machine, demonstrated an accuracy, precision, recall, and F1 score of 0.79, 0.82, 0.93, and 0.87, respectively. Radiomics features including the putamen and ventral diencephalon ranked as the most important for suggesting JME.
Conclusions: Radiomics models with multiple regions of interest in routine brain MRI could be an auxiliary tool in discerning patients with JME from HCs.
Funding: This research did not receive any grant from the public, commercial, or not-for-profit sector funding agencies.
Neuro Imaging