Related Experiment Video
Updated: Oct 16, 2025
08:52
Software-Assisted Quantitative Measurement of Osteoarthritic Subchondral Bone Thickness
Published on: March 18, 2022
3.2K
A deep learning method for predicting knee osteoarthritis radiographic progression from MRI
Jean-Baptiste Schiratti1, Rémy Dubois1, Paul Herent1
1Owkin, 12 Rue Martel, 75010, Paris, France.
Arthritis Research & Therapy
|October 19, 2021
Summary
A deep learning model can predict knee osteoarthritis (OA) structural progression using MRI scans, outperforming radiologists. This AI tool aids in identifying patients at high risk for disease modification drug development.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Orthopedics
Background:
- Accurate identification of knee osteoarthritis (OA) patients with rapid structural progression is crucial for developing effective disease-modifying drugs.
- Current methods for predicting OA progression require improvement to facilitate clinical trial enrollment and therapeutic development.
Purpose of the Study:
- To develop and evaluate a deep learning model for predicting structural knee osteoarthritis progression.
- To compare the performance of the deep learning model against trained radiologists in identifying patients at risk for cartilage degradation.
Main Methods:
- Utilized 9280 knee magnetic resonance (MR) images from 3268 patients in the Osteoarthritis Initiative (OAI) database.
- Implemented a deep learning classification model using MR images and clinical variables (e.g., BMI) to predict joint space narrowing at 12 months.
- Assessed model performance using Receiver Operating Characteristic Area Under the Curve (ROC AUC) scores.
Main Results:
- The deep learning model achieved a ROC AUC of 65% for predicting joint space narrowing, outperforming trained radiologists (58.7%).
- A separate model predicting pain progression (WOMAC index) achieved a ROC AUC of 72%.
- Attention maps highlighted specific regions of interest for predicting structural changes and pain.
Conclusions:
- Deep learning shows significant promise in analyzing knee OA progression from MR images.
- This AI approach has the potential to assist radiologists in identifying high-risk OA patients, aiding in drug development and clinical decision-making.
