Algorithms for analysis of knee radiographs

Value Proposition

Knee osteoarthritis accounts affects at least 19% of American adults aged 45 years and older. It is a disabling joint disease with causes that remain poorly understood but are commonly attributed to aging and obesity. The radiograph remains the most prevalent tool in the evaluation of the osteoarthritis joint. Algorithms that can evaluate X-rays for signs of joint disease will enhance existing radiologist workflows.


Researchers at Duke have developed an algorithm for characterization of knee X-ray images. The method uses machine learning to assign grading of osteoarthritis in the knee, evaluate osteophyte formation and joint space narrowing, and predict the extent of overall joint deterioration.

Right knee X-ray

Duke File (IDF) Number



  • Mazurowski, Maciej
  • Balu, Suresh
  • Hutyra, Carolyn
  • Jiranek, William "Bill"
  • Li, Nianyi
  • Mather, Richard "Chad"
  • O'Donnell, Jonathan
  • Said, Nicholas "Nick"

For more information please contact


School of Medicine (SOM)