A Progressive Risk Formulation for Enhanced Deep Learning based Total Knee Replacement Prediction in Knee...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Knee osteoarthritis is a leading cause of disability.
• Predicting the need for Total Knee Replacement (TKR) is crucial for patient management.
• Current prediction models often analyze scans independently, potentially missing disease progression dynamics.

Purpose of the Study:

• To develop deep learning models for predicting TKR need in knee osteoarthritis patients.
• To incorporate a novel progressive risk formulation for improved prediction accuracy using longitudinal data.
• To enable TKR prediction from single or multiple scans.

Main Methods:

• Developed deep learning models utilizing a dual-model risk constraint architecture.
• Trained models on knee radiographs and MRIs from the Osteoarthritis Initiative (OAI) and Multicenter Osteoarthritis Study (MOST).
• Enforced a progressive risk formulation constraint during training for patients with multiple scans to ensure risk stability or increase over time.

Main Results:

• The proposed models demonstrated superior performance compared to baseline conventional models.
• Achieved an AUROC of 0.87 and AUPRC of 0.47 for 1-year TKR prediction on the OAI radiograph test set.
• Outperformed baseline models on MOST radiograph and both OAI and MOST MRI test sets, showing consistent improvements.

Conclusions:

• Deep learning models with a progressive risk formulation can accurately predict TKR need in knee osteoarthritis.
• The novel approach enhances prediction by considering disease progression, outperforming conventional methods.
• This technology holds promise for improved patient management and surgical planning in osteoarthritis care.