Figure 3
From: Deep learning enables automated scoring of liver fibrosis stages
The architecture and of (A) deep learning via convolutional neuron networks (CNN) using pretrained 7-layered AlexNet and non-deep learning via (B) Artificial Neural Networks (ANN), (C) Multinomial Logistic Regression (MLR), (D) Support Vector Machines (SVM) and Random Forests (RF) for computer aided liver fibrosis scoring. SHG processed images were resized and duplicated to be used as input for deep learning-based algorithm, C1-C5: convolution layer, FC6-FC7: fully connected layer. Extracted morphological or/and textural features were used as input for non-deep learning-based algorithms.
