Unified CNN Approach for Fracture Detection Across All Anatomical Regions in Radiographic Images

Fractured bones are a serious health problem that can lead to serious consequences if not recognized and treated immediately. Early diagnosis of fractures is critical for efficient treatment and reducing long-term consequences on patient health. By incorporating AI-driven fracture detection technologies, healthcare may become more effective and efficient, supporting international efforts to provide universal access to fair and inclusive healthcare. This research employs a Bone Fracture multi-region X-ray dataset of radiographic scans to offer a complete fracture detection and categorization method utilizing a proposed Convolutional Neural Network (CNN) model. The collection contains 10,580 X-ray scans of broken and non-fractured bones from all anatomical locations, including the lower limb, upper limb, lumbar, hips, and knees. The proposed CNN model showed exceptional results, with an overall accuracy of 97.85%, and precision and recall rates of 98%. These results illustrate the usefulness of the model in reliably identifying and categorizing fractures across diverse anatomical locations, underlining its potential for increasing early fracture detection in clinical settings.