Fracture Detection in Pediatric Wrist Trauma X-ray Images Using YOLOv8 Algorithm

Hospital emergency departments frequently receive lots of bone fracturecases, with pediatric wrist trauma fracture accounting for the majority ofthem. Before pediatric surgeons perform surgery, they need to ask patients howthe fracture occurred and analyze the fracture situation by interpreting X-rayimages. The interpretation of X-ray images often requires a combination oftechniques from radiologists and surgeons, which requires time-consumingspecialized training. With the rise of deep learning in the field of computervision, network models applying for fracture detection has become an importantresearch topic. In this paper, we use data augmentation to improve the modelperformance of YOLOv8 algorithm (the latest version of You Only Look Once) on apediatric wrist trauma X-ray dataset (GRAZPEDWRI-DX), which is a publicdataset. The experimental results show that our model has reached thestate-of-the-art (SOTA) mean average precision (mAP 50). Specifically, mAP 50of our model is 0.638, which is significantly higher than the 0.634 and 0.636of the improved YOLOv7 and original YOLOv8 models. To enable surgeons to useour model for fracture detection on pediatric wrist trauma X-ray images, wehave designed the application "Fracture Detection Using YOLOv8 App" to assistsurgeons in diagnosing fractures, reducing the probability of error analysis,and providing more useful information for surgery.