Multi-View Digital Mammography Mass Classification: A Convolutional Neural Network Model Approach

Breast cancer (BC) is a disease in which malignant (cancer) cells form in the tissues of the breast. Among various breast screening techniques, mammography is the current gold standard for early detection of BC. In this paper, a two-stage breast mass classification model is innovatively proposed. In specific, stage-one is a feature extraction stage in which mammogram features are extracted and selected by a pre-trained convolutional neural network (CNN) model. Then, in stage-two, CNN-extracted mammogram features are classified as benign or malignant cases by a two-layered neural network classifier. Our proposed method has been evaluated on the subset of the INbreast database. The results show that our model obtains better performance than previous advanced classification methods.