Optimizing Wrist Fracture Detection: The Efficacy of YOLOv11 in Medical Imaging

Abstract

Wrist fractures pose significant health challenges to everyday functionality specially among extreme ages of populations, thereby necessitating timely and accurate detection. Our study evaluates the effectiveness of the You Only Look Once (YOLOv11) algorithm for automated wrist fracture detection using collected 600 X-ray dataset of images. Our proposed YOLOv11 model achieved a mean average precision (mAP@50) of 0.74) surpassing previous models such as YOLOv8 (mAP@50: 0.64) and YOLOv10 (mAP@50: 0.632) on the GRAZPEDWRI-DX dataset. The model also demonstrated superior localization of complicated fractures, with a 10% reduction in false negatives compared to baselines. Metal implants and overlapping fractures were identified as the challenges, suggesting the need for synthetic data and multi-view X-ray analysis. The study underscores the potential of YOLOv11 for real-time clinical applications, a potential for offering a robust framework for enhancing diagnostic accuracy in wrist fracture detection.

Ähnliche Arbeiten