Figure 3.

Receiver operating characteristic (ROC) plots showing the estimated area under curve (AUC) values obtained from the two classifiers. (a,b) A comparison of the performance by the two classifiers (k-NN and SVM) based on week 0 and week 4 time-points using all features. (c,d) The comparison of the performance of QUS + QUS-Tex¹ (without texture derivatives) and QUS + QUS-Tex¹ + QUS-Tex¹-Tex² (with texture derivatives) data set by k-NN and SVM models at week 4.