Related Experiment Video
Updated: Sep 16, 2025
06:31
Author Spotlight: Enhancing Rheumatoid Arthritis Research Through HR-pQCT Imaging Analysis
Published on: October 6, 2023
2.4K
Radiographic Bone Texture Analysis using Deep Learning Models for Early Rheumatoid Arthritis Diagnosis
Yun-Ju Huang1,2, Chiung-Hung Lin3,2, Shun Miao4
1Division of Rheumatology, Allergy and Immunology, Chang Gung Memorial Hospital, No.5, Fuxing St., Guishan District, Taoyuan, 333, Taiwan.
Journal of Imaging Informatics in Medicine
|July 7, 2025
Summary
Deep learning models can automatically analyze bone texture on hand radiographs to help diagnose early rheumatoid arthritis (RA). These AI tools show promise in identifying subtle changes, aiding clinicians in early RA detection and classification.
Area of Science:
- Radiology
- Artificial Intelligence
- Rheumatology
Background:
- Rheumatoid arthritis (RA) involves changes in bone microarchitecture (texture) around joints, which are difficult to detect radiographically.
- Early diagnosis of RA is crucial for effective treatment and management.
Purpose of the Study:
- To develop and validate deep learning models for quantitative assessment of periarticular texture from radiographs.
- To predict early rheumatoid arthritis (RA) diagnosis automatically, without human interpretation.
Main Methods:
- Two deep learning models, Deep Texture Encoding Network (Deep-TEN) and ResNet-50, were trained and validated on hand radiographs from early RA patients and non-RA controls.
- Models were evaluated for their ability to predict RA probability based on segmented periarticular texture features.
Main Results:
- The ResNet-50 model achieved a higher area under the curve (0.73) compared to Deep-TEN (0.69) in predicting RA.
- High texture scores from both models were significantly associated with increased odds of RA, with ResNet-50 showing higher predictive values.
- Deep learning models demonstrated moderate to strong associations with RA risk groups.
Conclusions:
- Fully automated quantitative assessment of periarticular texture using deep learning is feasible.
- These AI models can assist in the classification and early detection of rheumatoid arthritis.
- Deep learning offers a potential tool to enhance radiographic analysis for early RA diagnosis.
