Related Experiment Video
Updated: Jun 2, 2025
05:05
Four-Dimensional CT Analysis Using Sequential 3D-3D Registration
Published on: November 23, 2019
7.9K
Frontal plane mechanical leg alignment estimation from knee x-rays using deep learning
Kenneth Chen1,2, Christoph Stotter3, Christopher Lepenik4
1Department for Health Sciences, Medicine and Research, University of Continuing Education Krems, Krems, Austria.
Osteoarthritis and Cartilage Open
|January 15, 2025
Summary
A new deep learning model accurately identifies lower limb malalignment from knee X-rays. This method aids in selecting study participants and managing patients with knee osteoarthritis (OA).
Area of Science:
- Orthopedics
- Radiology
- Artificial Intelligence
Background:
- Lower limb malalignment is a significant factor in knee osteoarthritis (OA) progression and symptom severity.
- Accurate assessment of leg alignment is crucial for patient stratification in clinical studies and treatment planning.
Purpose of the Study:
- To develop and validate a deep learning model for classifying lower limb alignment from knee antero-posterior (AP)/postero-anterior (PA) radiographs.
- To assess the model's performance using adjustable hip-knee-ankle (HKA) angle thresholds.
Main Methods:
- Utilized a dataset of 8878 digital radiographs, including full-leg x-rays (LLRs) and knee x-rays.
- Employed a two-step validation process comparing model predictions on knee images to ground truth from LLRs.
Main Results:
- The deep learning model demonstrated high accuracy in classifying leg alignment.
- Sensitivity and specificity ranged from 0.74 to 0.92 across different thresholds and image types (with/without positioning frames).
Conclusions:
- The developed model accurately classifies lower limb malalignment using only knee radiographs.
- This approach offers a practical alternative to full leg radiographs (LLRs), improving precision in study selection and patient management.
