Research on Key Technologies of Human Upper Limb Bone Diagnosis based on Convolutional Neural Network

Fracture is a common disease in medical clinic, which is often diagnosed based on medical images. In order to reduce the reading time of experts and reduce the misdiagnosis rate, a medical auxiliary detection system for upper limb bone abnormality detection is proposed in this paper. The convolution neural network method is used to detect and diagnose the upper limb bone. In this paper, the open mura data set is used in the data set, and the adaptive contrast restriction algorithm is used to process the bone image, so that the bone edge is more obvious and the focus position is more clearly visible. Then, the upper limb bone injury data set and the upper limb bone position classification data set are constructed respectively. For medical image recognition, this paper proposes an IDNet network based on perception and Densenet: in the task of upper limb bone injury classification, the accuracy of IDNet reaches 0.92; In the classification task of upper limb bone parts, the accuracy of IDNet reached 0.99.