Machine learning-based analysis of multi-region bone fracture detection and classification using biomedical images

Bone fractures are a prevalent medical concern that necessitates precise and prompt diagnosis. Interpreting X-ray pictures to detect fractures is a complex and time-intensive task that requires specialized radiological expertise. This process is further complicated in under-resourced or high-demand clinical settings, when access to proficient radiologists is restricted and diagnostic inaccuracies may occur owing to exhaustion or excessive workload. These issues underscore the pressing necessity for automated, intelligent solutions to aid in diagnosing and classifying bone fractures. This study proposes a Deep Learning-Driven Computer Vision Model for Multi-Region Bone Fracture Detection and Classification (DLCVM-MBFDC) technique. The proposed method commences with a Wiener Filtering (WF) preparation phase to diminish image noise. Subsequently, DenseNet201 is utilized to extract deep features from biological pictures. A stacked autoencoder (SAE) model is employed for fracture identification and classification, while the wild gibbon optimizer algorithm (WGOA) model is utilized for hyperparameter optimization of the SAE model. The efficacy of the DLCVM-MBFDC approach is examined under the Kaggle dataset. The comparison study of the DLCVM-MBFDC approach portrayed a superior accuracy value of 97.85 % over existing models.