Based on digital medical image data, doctors and experts can observe and analyze pathological images through computers, thus giving diagnosis results. Based on the deep learning model, a computer aided diagnosis system targeted at various disease types has been developed. The main research contents include: Colon cancer pathological image detection system; Grade discrimination system of breast cancer pathological image; CT image-oriented pulmonary nodule benign and malignant judgment system; Thyroid nodule benign and malignant diagnosis system based on ultrasound image; Computer aided pneumoconiosis grade discriminant analysis system, etc. Among them, the benign and malignant detection accuracy of pathological images is 98%, and the benign and malignant detection accuracy of CT images is 86%. After communicating with relevant experts, we know that the accuracy has reached the judgment level of an very experienced doctor.