A Breast Image Classification Method based on Deep Convolutional Neural Network

Breast cancer is a serious threat to women's physical and mental health. Computer-based breast cancer assistant diagnosis system can help radiologists reduce workload, reduce missed diagnosis rate, and improve accuracy to a certain extent, but the performance still needs to be further improved to meet clinical needs. Aided diagnosis systems based on traditional machine learning and pattern recognition methods require manual design and feature extraction, which not only requires rich medical expertise, but also difficult to extract deep features of the image, causing a bottleneck in system performance. In this paper, a model based on deep convolutional neural network is designed and used for breast cancer assistant diagnosis. Experimental results show that the area under the ROC curve can reach 0.896, which has achieved good performance.