Classification of Radiologically Isolated Syndrome and Clinically Isolated Syndrome with Machine-Learning Techniques

Aug 4, 2026

Upload by Ivy

Machine-learning classification methods are applied to multimodal 3T MRI biomarkers to distinguish radiologically isolated syndrome (RIS) from clinically isolated syndrome (CIS), two early expressions along the multiple sclerosis spectrum. The study uses morphometric and diffusion-related measures from 17 RIS and 17 CIS patients for single-subject level prediction. The best models employ Naive Bayes, Bagging, and Multilayer Perceptron, achieving 78% accuracy. Three specific brain features drive performance: left rostral middle frontal gyrus volume and fractional anisotropy in the right amygdala and right lingual gyrus.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages24

Reading Time1 h 0 min

AccessFree to download

Abstract

Introduction

Methods

Results

Conclusions

Keywords

Classification of Radiologically Isolated Syndrome and Clinically Isolated Syndrome with Machine-Learning Techniques Virginia Mato-Abad, 1 Andrés Labiano-Fontcuberta,2 Santiago Rodríguez-Yáñez,1 Rafael García-Vázquez,1 Cristian R Munteanu,3, 4 Javier Andrade-Garda,1 Angela Domingo-Santos,2 Victoria Galán Sánchez-Seco,2 Yolanda Aladro,5 Mª Luisa Martínez-Ginés,6 Lucía Ayuso,7 Julián Benito-León,2, 8, 9 * From the ISLA, 1 Computer Science Faculty, A Coruna University, A Coruña, Spain; Department of Neurology,2 University Hospital “12 de Octubre”, Madrid, Spain; RNASA-IMEDIR,3 Computer Science Faculty, A Coruna University, A Coruña, Spain; Biomedical Research Institute of A Coruña (INIBIC)4 , University Hospital Complex of A Coruña (CHUAC), A Coruña, Spain; Department of Neurology,5 Getafe University Hospital, Getafe, Spain; Department of Neurology,6 University Hospital “Gregorio Marañón”, Madrid, Spain; Department of Neurology,7 University Hospital “Principe de Asturias”, Alcalá de Henares, Spain; Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED),8 Spain; Department of Medicine,9 Complutense University, Madrid, Spain. Abstract. Introduction: The unanticipated magnetic resonance imaging (MRI) detection in the brain of asymptomatic subjects of white matter lesions suggestive of multiple sclerosis (MS) has been named as radiologically isolated syndrome (RIS) . As the difference between early MS (i.e. , clinically isolated syndrome [CIS]) and RIS is the occurrence of a clinical event, it should be logical to improve detection of subclinical form without interfering with MRI as there are radiological diagnostic criteria for that. Our objective was to use machine-learning classification methods to identify morphometric measures that help to discern patients with RIS from those with CIS. Methods: We used a multimodal 3T MRI approach by combining MRI biomarkers (cortical thickness, cortical and subcortical grey matter volume, and white matter integrity) of a cohort of 17 RIS and 17 CIS patients for single-subject level classification. Results: The best proposed models to predict the CIS and RIS diagnosis were based on the Naive Bayes, Bagging and Multilayer Perceptron classifiers using only three features: the left rostral middle frontal gyrus volume, and the fractionalanisotropy values in the right amygdala and in the right lingual gyrus. The Naive Bayes obtained the highest accuracy (overall classification, 0.765 and AUROC, 0.782) . Conclusions: A machine-learning approach applied to multimodal MRI data may differentiate between the earliest clinical expressions of MS (CIS and RIS), with an accuracy of 78% . Keywords: Machine-learning, Magnetic Resonance Imaging, Diffusion Tensor Imaging, Radiologically Isolated Syndrome, Clinically Isolated Syndrome, Multiple Sclerosis, Naive Bayes Classifier, Multilayer Perceptron, Bagging. **Corresponding author: Julián Benito-Leó[n. Av. de](n. Av. de) la Constitución 73, portal 3, 7º Izquierda, 28821 Coslada, Madrid, [Spain. E-mail: jbenitol67@gmail.com](Spain. E-mail: jbenitol67@gmail.com) Word count abstract: 222; Number of references: 36; Number of tables: 2; Word count paper: 3,000 . Competing interests The authors declare no competing financial interests. Authors Roles: Dr. Virginia-Mato ([virginia.mato@udc.es](virginia.mato@udc.es)) collaborated in: 1) the conception, organization of the research project; 2) the statistical analysis design, and 3) the writing of the manuscript first draft and the review and critique of the manuscript. Dr. Labiano-Fontcuberta ([gandilabiano@hotmail.com](gandilabiano@hotmail.com)) collaborated in: 1) the conception, organization and execution of the research project; 2) and the review and critique of the manuscript. Dr. Rodríguez-Yáñez ([santiago.rodriguez@udc.es](santiago.rodriguez@udc.es)) collaborated in: 1) the conception, organization of the research project; and 2) the revi

FAQs

What is the goal of the machine-learning approach in this study?

To identify morphometric MRI measures that help distinguish patients with radiologically isolated syndrome (RIS) from those with clinically isolated syndrome (CIS).

How were MRI data collected and processed for classification?

The study uses multimodal 3T MRI biomarkers, combining cortical thickness, grey matter volumes, and white matter integrity measures for single-subject level classification.

Which models and MRI features performed best for RIS versus CIS prediction?

Naive Bayes with three features achieved the highest accuracy, using left rostral middle frontal gyrus volume and fractional anisotropy values in the right amygdala and right lingual gyrus.

Aug 4, 2026

Upload by Ivy