Sjögren's disease & overlap diseases
Deep learning in the precise assessment of primary Sjögren’s syndrome based on ultrasound images
Authors' conclusion
The DL model based on greyscale US images showed diagnostic potential in the precise assessment of pSS patients in the SMGs, PGs and LGs, outperforming conventional radiologist evaluation.
Study context
- Population
- Primary Sjögren's syndrome (pSS) patients diagnosed by 2016 ACR/EULAR criteria and sex- and age-matched healthy controlsdevelopment: 72 pSS + 72 controls
- Comparison
- Deep learning (DL) model based on greyscale ultrasound (US) images vs two radiologists
- Outcome
- Diagnostic performance (AUC) of the DL model for pSS
Methodological detailsmulticentre, prospective, sex- and age-matched healthy controls, 864 and 164 greyscale US images collected for development and validation
- Multicentre
- Prospective
- Sex- and age-matched healthy controls
- 864 and 164 greyscale US images collected for development and validation
Performance
As reported in the abstract
AUC of DL model, model cohort
| Test or model | AUC, submandibular glands (SMGs) | AUC, parotid glands (PGs) | AUC, lacrimal glands (LGs) |
|---|---|---|---|
| DL model | 0.92 | 0.93 | 0.91 |
DL model
AUC, submandibular glands (SMGs)0.92
AUC, parotid glands (PGs)0.93
AUC, lacrimal glands (LGs)0.91
Further findings
- AUC of DL model, validation cohort
- 0.90 · AUC, SMGs · AUC, PGs 0.88 · AUC, LGs 0.87
Calibration
- Prediction probability consistent with actual probability in model and validation cohorts
Original publication
Deep learning in the precise assessment of primary Sjögren’s syndrome based on ultrasound images. Rheumatology (Oxford). Published 3 June 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →