Design of integrated interactive system for pre-diagnosis of breast cancer pathological images based on CNN and PyQt5

In today's society, breast cancer is becoming more and more serious, and the promotion of related medical means is imminent. They require modern technology support led by artificial intelligence, such as computer-aided diagnosis system for breast cancer. The existing computer-aided diagnostic systems of breast cancer have high cost, high requirements for the configuration of hospital hardware facilities, and are basically limited to the classification of benign and malignant. The paper proposes L_MiniVGGNet (improved VGGNet) and T_MobileNetV2 (improved MobileNetV2) algorithms, and a simple integrated breast cancer pathologic image pre-classification system. The system uses L_MiniVGGNet and T_MobileNetV2 as the core algorithms to solve the problem of binary and quaternary classification of pathological images, respectively. It adapts to both tensorflow and pytorch frameworks. In this paper, BreakHis and BACH open data sets are used, and data enhancement and transfer learning algorithms are combined to train high-precision and low-memory models and save them. PyQt5 library is used to build human–computer interaction GUI system interface, and images are imported according to the path required by system users and diagnosis results are predicted.