Related Experiment Video
Updated: Jan 12, 2026
07:25
Author Spotlight: A Focus on Standardized Salivary Gland Ultrasound Protocol in Connective Tissue Disease Research
Published on: October 13, 2023
4.9K
An Ultrasound-Based Machine Learning Model for Differentiating IgG4-Related Sialadenitis and Primary Sjögren's
Huan-Zhong Su1, Long-Cheng Hong1, Yu-Hui Wu1
1Department of Ultrasound, The First Affiliated Hospital of Xiamen University, School of Medicine, Xiamen University, Xiamen, China.
International Dental Journal
|November 7, 2025
Summary
An interpretable machine learning model using ultrasound scores accurately differentiates immunoglobulin G4-related sialadenitis (IgG4-RS) from primary Sjögren's syndrome (pSS), offering a non-invasive diagnostic tool.
Area of Science:
- Rheumatology
- Medical Imaging
- Machine Learning
Background:
- Distinguishing immunoglobulin G4-related sialadenitis (IgG4-RS) from primary Sjögren's syndrome (pSS) is clinically challenging.
- Ultrasound (US) findings are crucial in evaluating salivary gland diseases.
Purpose of the Study:
- To develop and validate an interpretable machine learning (ML) model for differentiating IgG4-RS from pSS using US scoring.
- To identify key US features that predict IgG4-RS or pSS.
Main Methods:
- A cohort of 263 patients with IgG4-RS or pSS was divided into training and validation sets.
- Least absolute shrinkage and selection operator (LASSO) regression identified predictive features, and various ML models were trained.
- Model performance was evaluated using AUC, accuracy, sensitivity, and specificity; feature importance was assessed with SHAP analysis.
Main Results:
- The light gradient boosting machine model achieved an AUC of 0.955 in the validation cohort.
- Key differentiating features included sex, dry mouth/eyes, gland enlargement, and US scores (PGUS, SMGUS, SMGV).
- SHAP analysis identified specific US features and clinical symptoms predictive of pSS.
Conclusions:
- An interpretable ML model utilizing US scoring provides an accurate and non-invasive method for differentiating IgG4-RS from pSS.
- This approach enhances diagnostic capabilities for salivary gland diseases.
