Classification of Wrist Fracture from X-Ray Images Using DenseNet-121

Wrist Bone fractures represent a significant challenge in healthcare, with millions of cases managed globally by medical professionals each year. Accurately diagnosing fractures is often complex, and the process of determining whether a bone is fractured remains time-consuming and labor-intensive. However, in today’s technologically advanced era, such tasks can potentially be automated using deep learning models. Among these models, this study focuses on the Convolutional Neural Network (CNN) architecture known as DenseNet121. The aim of this research is to assess the effectiveness of DenseNet121 in classifying data to determine whether a bone is fractured or intact. A publicly available dataset has been used to train the model, followed by a 4-fold cross-validation to evaluate its performance. The model has demonstrated an impressive overall accuracy of 96.25%. This result highlights the potential of DenseNet121 in significantly reducing the diagnostic workload for medical practitioners, while also providing more accurate and efficient identification of fractures. The findings of this study underscore the value of leveraging deep learning technologies to support clinical decision-making and improve diagnostic accuracy in the field of fracture detection.