Scholarlink

<p>This systematic review offers novel insights into the application of deep learning (DL) techniques across a broad spectrum of orthopedic disorders—namely osteoarthritis (OA), rheumatoid arthritis (RA), osteoporosis, and scoliosis the top contributors to years lived with disability (YLDs) affecting 1.63&#xa0;billion individuals globally as of 2020. Early-stage diagnosis is often complicated by unclear and overlapping imaging symptoms and subtle variations, but advance DL models are increasingly capable of overcoming through automated and precise classification. Following PRISMA guidelines, research articles published between 2019 and 2024 were retrieved from databases including Scopus, PubMed, Google Scholar, and IEEE. A total of 118 relevant studies were selected based on inclusion criteria focused on DL-based models for orthopedic classification. This review distinctively emphasizes on critical role of preprocessing pipelines covering image augmentation, normalization, segmentation and fusion techniques for model performance and generalizability. It highlights various DL model architectures (including CNNs, pretrained models, hybrid frameworks, and transformers) used across multiple orthopedic diagnoses. The study bridges the gap between technical modeling and clinical interpretability by evaluating the use of explainable AI techniques among screened articles. The findings demonstrate that DL models contribute effectively to distinguishing between different severity levels and overlapping cases in orthopedic conditions. By synthesizing insights from recent research, this review supports clinicians and researchers in identifying effective DL approaches and addresses ongoing challenges like dataset heterogeneity, hardware requirements, and the challenges of real-world deployment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deep Learning Techniques for the Diagnosis and Detection of Orthopedic Conditions: A Systematic Review of Recent Advances and Challenges

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • Poojan Shah,
  • Yogesh Kumar,
  • Tanmay Bhowmik

摘要

This systematic review offers novel insights into the application of deep learning (DL) techniques across a broad spectrum of orthopedic disorders—namely osteoarthritis (OA), rheumatoid arthritis (RA), osteoporosis, and scoliosis the top contributors to years lived with disability (YLDs) affecting 1.63 billion individuals globally as of 2020. Early-stage diagnosis is often complicated by unclear and overlapping imaging symptoms and subtle variations, but advance DL models are increasingly capable of overcoming through automated and precise classification. Following PRISMA guidelines, research articles published between 2019 and 2024 were retrieved from databases including Scopus, PubMed, Google Scholar, and IEEE. A total of 118 relevant studies were selected based on inclusion criteria focused on DL-based models for orthopedic classification. This review distinctively emphasizes on critical role of preprocessing pipelines covering image augmentation, normalization, segmentation and fusion techniques for model performance and generalizability. It highlights various DL model architectures (including CNNs, pretrained models, hybrid frameworks, and transformers) used across multiple orthopedic diagnoses. The study bridges the gap between technical modeling and clinical interpretability by evaluating the use of explainable AI techniques among screened articles. The findings demonstrate that DL models contribute effectively to distinguishing between different severity levels and overlapping cases in orthopedic conditions. By synthesizing insights from recent research, this review supports clinicians and researchers in identifying effective DL approaches and addresses ongoing challenges like dataset heterogeneity, hardware requirements, and the challenges of real-world deployment.