AI Pathology Detection and Osteoarthritis Grading on Knee X-Rays
A multi-site Indian study of AI that reads knee X-rays — detecting joint space narrowing, sclerosis, osteophytes and other pathologies and grading osteoarthritis severity — trained on 1.3 million radiographs from government, private and SME hospitals.
Key findings
- Trained on 1.3 million knee X-rays from a multi-site Indian clinical trial across government, private and SME hospitals
- Detects joint space narrowing, sclerosis, osteophytes, tibial spikes, alignment issues and soft tissue anomalies
- Grades osteoarthritis severity with pathology-specific models (ResNet15 for joint space narrowing, DenseNet for grading)
- Subgroup analyses across age, gender and manufacturer support generalizability
Plain-English summary
- The question
- Osteoarthritis is often diagnosed late. Can AI read a knee X-ray, pick out the key pathologies and grade osteoarthritis severity across very different hospitals and machines?
- The data
- 1.3 million knee X-rays from a multi-site Indian clinical trial spanning government, private and SME hospitals, chosen for diversity in demographics, imaging equipment and clinical settings.
- The method
- Pathology-specific deep learning models — for example ResNet15 for joint space narrowing and DenseNet for osteoarthritis grading — evaluated with mean squared error, intersection over union and Dice coefficient.
- The result
- The authors report strong precision, recall and negative predictive value across diverse imaging environments, and subgroup analyses by age, gender and manufacturer confirmed generalizability. The published summary does not give headline percentages.
Why it matters for hospitals and radiologists
The paper frames consistent osteoarthritis grading as the route to timely, personalized treatment. A model validated across government, private and smaller hospitals reflects the mix of settings Indian radiology actually runs in, which is what makes the result usable beyond a single center.
Related at 5C: Bionic, the AI co-radiologist
Citation
Bargava Subramanian, Naveen Kumarasami, Praveen Shastry, Kalyan Sivasailam, Anandakumar D, Keerthana R, Mounigasri M, Abilaasha G, Kishore Prasath Venkatesh. A Multi-Site Study on AI-Driven Pathology Detection and Osteoarthritis Grading from Knee X-Ray. arXiv preprint arXiv:2503.22176, 2025. https://doi.org/10.48550/arXiv.2503.22176
