Research on rib fracture auxiliary diagnosis based on convolutional neural network

Abstract

Rib fractures are a common condition in chest injuries, which may be caused by traffic accidents, falls from heights, and daily exercise. Traditional detection methods such as CT scanning are prone to high rates of missed and false detections due to the complexity of rib fractures. However, detection methods based on deep learning technology have better detection performance and speed advantages. This article uses a network model called YOLOv12+Mamba for rib fracture detection, and compares it with other common detection algorithms through experiments. The test results show that the YOLOv12+Mamba network significantly improves the average accuracy (mAP_0.5), and its accuracy and recall also exceed other models, proving the effectiveness of this network. On the basis of the YOLOv12 model, this article has improved the problems and developed a new Mamba model. The outstanding performance of this model in image recognition comes from its efficient object detection and classification capabilities, as well as its support for faster training and inference processes, resulting in outstanding performance when processing large-scale image data. The improvement plan increases mAP_0.5 from 0.8612 to 0.9354, mIoU by 0.41% -1.2%, and inference speed by only 3%. Compared with the initial model, the improved prediction results are significantly better. The YOLOv12+Mamba network proposed in this article significantly improves the accuracy of rib fracture detection and effectively reduces the workload of doctors. This model provides a satisfactory solution to the problems of missed and false detections in traditional detection methods, improving diagnostic efficiency and accuracy.