Deep Learning-based Alignment Measurement in Knee Radiographs

2025-06-22

AI Summary

Clinical assessment of knee alignment from anteroposterior (AP) radiographs relies on manual measurements requiring full-length lower-limb radiographs—a time-consuming, operator-dependent process with limited reproducibility. Method: We propose the first deep learning–based fully automated radiographic knee alignment analysis framework. Our method employs an attention-gated hourglass network to simultaneously localize over 100 anatomical landmarks and reconstruct complete joint morphology, directly estimating the tibiofemoral angle (TFA). Contribution/Results: In clinical validation, our method achieves a mean absolute error of 1.0° against ground-truth manual measurements. Intraclass correlation coefficients (ICCs) reach 0.97 (preoperative) and 0.86 (postoperative), demonstrating high accuracy and clinical reliability. The framework eliminates manual annotation and full-leg imaging requirements, significantly improving efficiency in preoperative planning and postoperative evaluation. It establishes a practical, quantitative paradigm for knee radiographic analysis.

Technology Category

• Machine Learning: Kernel Methods
• Computer Vision: Other Foundations of Computer Vision
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Security and Privacy: Large-scale security measurements
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Radiographic knee alignment (KA) measurement is important for predicting joint health and surgical outcomes after total knee replacement. Traditional methods for KA measurements are manual, time-consuming and require long-leg radiographs. This study proposes a deep learning-based method to measure KA in anteroposterior knee radiographs via automatically localized knee anatomical landmarks. Our method builds on hourglass networks and incorporates an attention gate structure to enhance robustness and focus on key anatomical features. To our knowledge, this is the first deep learning-based method to localize over 100 knee anatomical landmarks to fully outline the knee shape while integrating KA measurements on both pre-operative and post-operative images. It provides highly accurate and reliable anatomical varus/valgus KA measurements using the anatomical tibiofemoral angle, achieving mean absolute differences ~1° when compared to clinical ground truth measurements. Agreement between automated and clinical measurements was excellent pre-operatively (intra-class correlation coefficient (ICC) = 0.97) and good post-operatively (ICC = 0.86). Our findings demonstrate that KA assessment can be automated with high accuracy, creating opportunities for digitally enhanced clinical workflows.

Problem

• Automates knee alignment measurement using deep learning
• Replaces manual, time-consuming traditional radiographic methods
• Enhances accuracy in pre- and post-operative assessments

Innovation

• Deep learning for knee alignment measurement
• Hourglass networks with attention gates
• Localizes 100+ landmarks for knee shape