• Abdomen Computed Tomography-based Radiomics and Machine Learning for Bone Mineral Density Evaluation: A review
  • Mahmoud Mohammadi-Sadr,1,* Amirreza Sadeghinasab,2 Fatemeh Mazaheri,3 Mohammadreza Elhaie,4
    1. Department of Medical Physics, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran
    2. Department of Medical Imaging and Radiation Sciences, School of Allied Medical Sciences, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran
    3. Medical Physics and Biomedical Engineering Department, School of Medicine, Tehran University of Medical Sciences, Tehran, Iran
    4. Department of Medical Physics, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran


  • Introduction: The evaluation of bone mineral density (BMD) is crucial for diagnosing and managing osteoporosis and other metabolic bone diseases. Traditional methods, such as dual-energy X-ray absorptiometry (DXA), have limitations in accessibility and precision. Recent advancements in radiomics and machine learning (ML) offer promising alternatives. This review explores the application of abdomen computed tomography (CT)-based radiomics combined with ML techniques for BMD evaluation, highlighting their potential to enhance diagnostic accuracy and clinical outcomes.
  • Methods: A comprehensive literature search was conducted across multiple databases, including PubMed and Scopus focusing on studies published between 2010 and 2023. Keywords included “abdominal CT,” “radiomics,” “machine learning,” and “bone mineral density.” Selected studies were assessed for methodological quality and relevance. Data extraction focused on the type of radiomics features used, ML algorithms applied, and the performance metrics reported. Comparative analyses were performed to evaluate the efficacy of CT-based radiomics and ML models against traditional DXA measurements.
  • Results: The review identified 25 relevant studies that utilized abdomen CT-based radiomics and ML for BMD evaluation. Commonly extracted radiomics features included texture, shape, and intensity-based metrics. ML algorithms such as support vector machines (SVM), random forests, and deep learning models demonstrated high predictive accuracy, with some studies reporting area under the curve (AUC) values exceeding 0.90. Comparative analyses indicated that CT-based radiomics and ML models provided comparable or superior BMD assessments relative to DXA, particularly in detecting early-stage osteoporosis and assessing fracture risk.
  • Conclusion: Abdomen CT-based radiomics combined with ML techniques represents a promising approach for BMD evaluation. The integration of advanced imaging features and sophisticated ML algorithms can potentially overcome the limitations of traditional methods, offering enhanced diagnostic precision and early detection capabilities. Future research should focus on standardizing radiomics feature extraction, optimizing ML models, and validating these approaches in larger, diverse populations to facilitate clinical translation and widespread adoption.
  • Keywords: Bone Mineral Density, Computed Tomography, Radiomics, Machine Learning