YOLOv7-Based Approach for Detecting Hand and Forearm Bone Fractures in Radiology

Accurate detection of bone fractures is crucial for timely medical intervention and improved patient outcomes. Traditional diagnostic methods often rely on manual examinations, which can lead to delays and inaccuracies. This study explores the application of YOLOv7 for automated hand and forearm bone fracture detection. The research includes a detailed process of dataset collection, preprocessing, model training, and evaluation to assess YOLOv7's effectiveness in fracture identification. Experimental results demonstrate that extending the training to 800 epochs significantly improves detection performance, achieving a precision of 0.957, a recall of 0.554, and an mAP@0.5 of 0.644, compared to 300 epochs, which yielded a precision of 0.848, a recall of 0.442, and a mAP@0.5 of 0.409. These findings underscore the potential of YOLOv7 to enhance diagnostic accuracy and efficiency in clinical practice, ultimately benefiting patient care. The proposed approach highlights the transformative role of automated detection systems in advancing healthcare delivery and supporting clinicians with fracture diagnosis.