深度学习联合影像组学在骨骼肌肉疾病的研究进展OA

Musculoskeletal diseases are among the most prevalent and debilitating chronic conditions worldwide.With the acceleration of population aging,their incidence continues to rise,posing a major public health challenge.Conventional imaging-based diagnosis relies heavily on the subjective interpretation of clinicians and is limited by high inter-observer variability,insufficient sensitivity for early lesions,and a lack of robust quantitative assessment tools,making it difficult to meet the requirements of precision medicine.In recent years,the rapid development of deep learning and radiomics has provided new technical pathways for intelligent assessment and decision-making in musculoskeletal disorders.This review systematically summarizes the research progress of deep learning and radiomics in a range of musculoskeletal conditions,including osteoarthritis,osteoporosis and fragility fractures,bone tumors and benign-malignant differentiation,muscle diseases and muscle atrophy,as well as tendon and ligament injuries.We focus on their applications in automatic segmentation,computer-aided diagnosis,disease classification,progression prediction,treatment decision support,and prognostic evaluation,highlighting their potential advantages in improving diagnostic accuracy,enabling quantitative characterization of lesions,and supporting individualized therapeutic strategies.In addition,we outline the major challenges currently limiting clinical translation,such as insufficient data standardization,limited model interpretability,suboptimal multicenter generalizability,and uncertainties in implementation pathways.Finally,future research directions are discussed with the aim of providing methodological reference and theoretical support for early diagnosis,prognostic assessment,and precision treatment of musculoskeletal diseases based on deep learning and radiomics.