Fig. 6From: Richer fusion network for breast cancer classification based on multimodal dataROC curves and AUC area were compared integrating different convolutional layers of VGG16. Except for the different, the other experimental settings were controlled as follows: Structured data (denoising autoencoder, 580D, 0.2 missing rate, L1 regularization) + Fully connected layers (500, 100, 2)Back to article page