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

Mato-Abad, VirginiaLabiano-Fontcuberta, AndrésRodríguez-Yáñez, S.García Vázquez, RafaelMunteanu, Cristian-RobertAndrade-Garda, JavierDomingo-Santos, ÁngelaGalán Sánchez-Seco, VictoriaAladro, YolandaMartínez-Ginés, M. LuisaAyuso, LucíaBenito-León, Julián2023-11-212023-11-212019Mato-Abad, V., Labiano-Fontcuberta, A., Rodríguez-Yáñez, S., García-Vázquez, R., Munteanu, C.R., Andrade-Garda, J., Domingo-Santos, A., Galán Sánchez-Seco, V., Aladro, Y., Martínez-Ginés, M.L., Ayuso, L. and Benito-León, J. (2019), Classification of radiologically isolated syndrome and clinically isolated syndrome with machine-learning techniques. Eur J Neurol, 26: 1000-1005. https://doi.org/10.1111/ene.13923http://hdl.handle.net/2183/34309Versión final aceptada de: https://doi.org/10.1111/ene.13923[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 fractional anisotropy 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%.engTodos os dereitos reservados. All rights reserved.Machine learningMagnetic resonance imagingDiffusion tensor imagingRadiologically isolated syndromeClinically isolated syndromeMultiple sclerosisNaive Bayes classifierMultilayer perceptronBaggingClassification of Radiologically Isolated Syndrome and Clinically Isolated Syndrome with Machine-Learning Techniquesjournal articleopen access10.1111/ene.13923