Study on AO classification of distal radius fractures based on multi-feature fusion

Abstract Accurate classification of distal radius fracture is of great significance for improving the accuracy and success rate of subsequent bone-setting techniques. Based on the existing clinical distal radius fracture cases of the research group, this paper proposes an AO classification method based on distal radius fracture images based on multi-feature fusion for the problems of poor single feature expression and low accuracy of fracture classification by traditional classifiers and deep learning models. The fusion model uses two traditional features and the depth features extracted by AlexNet. After reducing the dimensions of the above features, a special neural network is designed to effectively fuse the reduced feature vectors. Finally, use the image retrieval classification scheme DML-K proposed in this paper to realize the specific classification of DRF images with the fused feature vectors. The experimental results show that the accuracy of the diagnostic model proposed in this paper on the distal radius fracture data set reaches 83.4%, and the F1 value reaches 0.815. Through horizontal and vertical comparison with other algorithms, the accuracy of the DRF diagnosis model proposed in this paper is increased by 5%, and the F1 value is increased by about 0.2, which fully verifies the effectiveness and feasibility of the method proposed in this study, and is expected to be applied Machine-aided diagnosis of distal radius fracture.