Advancements in deep learning-based image screening for orthopedic conditions: Emphasis on osteoporosis, osteoarthritis, and bone tumors

Abstract

Artificial intelligence (AI) has garnered increasing attention in the medical field. As the core technology of AI, deep learning (DL) has been extensively applied to the imaging-based screening of orthopedic diseases, primarily including image classification, segmentation, and risk prediction. This review systematically summarizes recent research advances, methodologies, and clinical applications of AI-assisted diagnostic technologies in orthopedic imaging, highlighting the practical value and development trends of DL in this field. By retrieving literature published over the past five years in PubMed and the Web of Science Core Collection, this study emphasizes the application of DL-based techniques in the screening of orthopedic conditions, such as osteoarthritis (OA), osteoporosis (OP), and bone tumors. The results demonstrate that DL-based methods exhibit excellent diagnostic performance and considerable clinical potential. However, despite the rapid increase in research output, there are still several challenges in this field, including the lack of high-quality datasets, the limited cross-institutional generalizability of models, the absence of standardized quality control protocols, and the urgent demand for multicenter clinical validation. Overall, DL holds great promise for enhancing diagnostic accuracy and improving patient outcomes in orthopedic imaging.