Breast Cancer Computer-Aided-Diagnosis System Based On Transfer Learning And Convolutional Neural Networks

Breast cancer Computer-aided detection(CAD)system is playing a more and more important role in medical detection and diagnosis.In order to classify tumor and nontumor in Magnetic Resonance Imaging(MRI),a novel breast cancer CAD system based on deep learning and transfer learning is designed.First,balance the imbalanced data sets and use data augmentation to deal with it.Then,use convolutional neural network(CNN)to extract CNN features from MRI data sets,use the same support vector machine(SVM)to evaluate the feature extraction abilities of different layers,select the highest F1 score layer as the node of fine-tuning,the layers behind it,which has relatively low dimension as the node of connection of new networks.Next,connect two newly designed fully-connected layers after the second node is selected.At last,freeze the layers before the node of fine-tuning,while other layers could be trained in the fine-tuning procedure.The CAD systems are built on three CNN networks,include VGG16,InceptionV3 and ResNet50.The effect of the system based on VGG16 and ResNet50 have the best performance.Then further exploration is made on the performance of twice transfer learning based on VGG16 network,and the result shows that twice transfer learning could improve the performance of the system.