Optimizing Multi-Region Fracture Detection Through AlexNet Deep Learning Architecture

The freedom of motion and quality of life of an individual can be greatly affected by common injuries like fractures. Early fracture detection is essential for efficient and timely treatment, reducing complications and improving recovery results. The X-ray image is used by the doctors to diagnose the fractured bone. The manual fracture identification method takes a long time and has a significant mistake probability. As a result, an automated technique for identifying fractured bones must be developed. To distinguish between healthy and fractured bone, the current study proposed a transfer learning-based neural network model. This paper uses an optimized AlexNet model to propose a comprehensive method for multi-region fracture diagnosis. The dataset that was used includes X-ray pictures of both broken and non-fractured bones from a variety of anatomical areas, such as the knees, hips, lumbar, lower limb, and upper limb. To discriminate between X-ray pictures that are fractured and those that are not, the AlexNet convolutional neural network—which is tuned for binary classification—is used in the process. With a 97.12% classification accuracy, the refined AlexNet model showed impressive accuracy. The model's high accuracy highlights its potential to help radiologists and other healthcare practitioners identify fractures early on, which might lead to better patient outcomes and more effective medical workflows.