Introducing radiomics model to predict active plaque in multiple sclerosis patients using magnetic resonance images

Abstract

Multiple Sclerosis (MS) is the most common non-traumatic disabling disease in young people. The prediction active plaque has the potential to offer new biomarkers for assessing the activity of MS disease. Consequently it supports patient management in the clinical setting and trials. This study aims to investigate the predictive capability of radiomics features for identifying active plaques in these patients using T2 FLAIR (Fluid Attenuated Inversion Recovery) images. For this purpose, a dataset images from 82 patients with 122 lesions was analyzed. Feature selection was performed using the Least Absolute Shrinkage and Selection Operator (LASSO) method. Six different classifier algorithms, namely K-Nearest Neighbors (KNN), Logistic Regression (LR), Decision Tree (DT), Support Vector Machine (SVM), Naive Bayes (NB), and Random Forest (RF), were employed for modeling. The models were evaluated using 5-fold cross-validation, and performance metrics including sensitivity, specificity, accuracy, area under the curve (AUC), and mean squared error were computed. A total of 107 radiomics features were extracted for each lesion, and 11 robust features were identified through the feature selection process. These features consisted of four shape features (elongation, flatness, major axis length, mesh volume), one first-order feature (energy), one Gray Level Co-occurrence Matrix feature (correlation), two Gray Level Run Length Matrix features (gray level non-uniformity, gray level non-uniformity normalized), and three Gray Level Size Zone Matrix features (low gray level zone emphasis, size zone non-uniformity, small area low gray level emphasis). The NB classifier demonstrated the best performance with an AUC, sensitivity, and specificity of 0.85, 0.82, and 0.66, respectively. The findings indicate the potential of radiomics features in predicting active MS plaques in T2 FLAIR images.