Figure 4
From: Deep learning enables automated scoring of liver fibrosis stages
The area under receiver operating characteristic (AUROC) values for various classification models from AlexNet-Convolutional Neuron Networks (CNN), conventional Artificial Neural Networks (ANN), non-linear Multinomial Logistic Regression (MLR), linear Support Vector Machines (SVM) and feature-ranking based Random Forests (RF) algorithms. AUROC was evaluated for CNN and ANN, MLR, SVM and RF with (A) morphological features (Features 1–21), (B) textural features (Features 21–130) and (C) both morphological and textural features (Feature 1–130). N.S.: non-significant difference, *adjusted p value is less than 0.005.
