The accuracy of the supervised learning models is roughly similar when using Input A or Input C but is decreased using Input B.

(A), (B), (C), (D) and (E) display the normalized confusion matrices for the logistic regression (LR), support vector machine (SVM), decision tree (DT), random forest (RF), and gradient-boosted decision tree (GBDT) models, respectively. Each matrix shows the performance on the same selected instance of cross-validation for models trained with Inputs A, B, and C from left to right. The greatest difficulty the models faced was in distinguishing early arterial phase (E-AP) from late arterial phase (L-AP) and late arterial phase (L-AP) from portal venous phase (PVP). Models trained using Input B have significantly greater difficulty correctly identifying E-AP than the models trained on the other two inputs.

(PDF)