Revolutionizing Bone Fracture Diagnosis: A Deep Learning Approach to X-Ray Image Analysis

Bones are the very important part of the human body, which provides body structure and protection for the internal organs. A bone fracture is a widespread scenario for the human body, which can lead to serious complications. Misdiagnosis of fractures is the most common mistake, resulting in treatment delays and permanent impairment. So, timely and accurate fracture detection is critical for proper treatment planning and medical diagnosis. X-ray imaging is a widely used diagnostic tool since manual interpretation is prone to errors. This study proposes an AI enhanced bone detection framework utilizing a bone fracture dataset consisting of 9463 X-ray images of fractured and non-fractured cases. Different preprocessing and data augmentation techniques played a major role in improving the dataset diversity and generalizability. The proposed methodology employs ResNet 50 for the feature extraction, enhancing it with the Bottleneck Attention Module (BAM) with dual attention strategies to refine critical features for effective fracture detection. With an accuracy of 97%, 96.12% precision, recall of 96.70%, and 96.38% F1 score, the suggested model outperformed other models like YOLOv8, Ensemble Model, ResNet50-DenseNet 121, and CNN. The results demonstrated that with improved feature representation and accuracy in bone fracture detection, the proposed model exhibits a valuable tool for enhanced patient care through early intervention and accurate fracture diagnosis.