WRIST FRACTURE PREDICTION USING TRANSFER LEARNING, A CASE STUDY
Rabia Javed, Tahir Abbas, Jamshaid Iqbal Janjua, Muhammad Abubakar Muhammad, Sadaqat Ali Ramay, Maryam Basit
Abstract
Patients regularly present to emergency rooms with suspected fractures, which frequently results in delayed treatment and poor recovery because fractures in images from X-rays are missed. Sometimes emergency care professionals who lack orthopaedic knowledge interpret these images incorrectly. Convolutional neural networks (CNNs), in particular, have attained human-level accuracy in classifying bone fractures. A deep learning model might reduce time wastage and incorrect diagnosis if it is created effectively. ResNet-101, a cutting-edge deep convolutional neural network frequently utilized in computer vision applications like object detection, image classification, and image segmentation, is used in our suggested framework, a WFP-TL model, which achieves 98.45% accuracy. ResNet-101, sporting 101 layers, regularly produces better results on benchmark datasets, MATLAB2020a is used for results and simulations. By teaching generalist medical practitioners on the front lines of healthcare, transfer learning improves patient care. The study demonstrates the potential of DL-based wrist fracture diagnosis on clinical radiographs, will serve as a foundation for future studies incorporating multi-view data for fracture classification.