Skip to main content
Fig. 2 | Journal of Translational Medicine

Fig. 2

From: Multi-parametric MRI phenotype with trustworthy machine learning for differentiating CNS demyelinating diseases

Fig. 2

Results of Multi-level Feature Selection and phenotype testing. AUC area under the curve, RSD relative standard deviation, MPR magnetization-prepared rapid gradient-echo. A Results of univariate feature selection and analysis. Top six T2 and T1-MPRAGE features and four clinical factors were selected and ranked according to AUC. The asterisk (*) represents statistical significance (p < 0.05) with Wilcoxon test. B Results of multivariate feature selection. The algorithm selected a subset of features having the highest AUC. The arrows indicate the stoping points where the highest AUC was achieved. C Comparison of diagnostic performance of T2, T1-MPRAGE, clinical phenotypes and the multi-parametric phenotype in the independent testing

Back to article page