Spondyloarthritis & psoriatic arthritis
Automated detection of active and structural magnetic resonance imaging (MRI) lesions in the sacroiliac joints in axial spondyloarthritis: training and validation across 3 phase III clinical trial datasets
Authors' conclusion
A fully automated DL-based approach can reliably detect active and structural SIJ MRI lesions in axSpA with robust external validation
Study context
- Population
- Patients with axial spondyloarthritis (axSpA) in phase III clinical trial datasets (MEASURE 1, PREVENT, SURPASS); MRI of the sacroiliac joints (SIJs)training 132
- Comparison
- Fully automated deep learning (DL) system vs expert reader evaluations
- Outcome
- Performance of automated detection of active (BMO) and structural (erosions, fat lesions, sclerosis, ankylosis) SIJ MRI lesions: AUC, balanced accuracy, sensitivity, specificity, kappa
Methodological detailsdeep learning, 2-stage automated pipeline, external validation on 2 independent datasets
- Deep learning
- 2-stage automated pipeline
- External validation on 2 independent datasets
Performance
As reported in the abstract
Automated SIJ lesion detection vs expert evaluations
| Test or model | Ankylosis AUC | Ankylosis balanced accuracy |
|---|---|---|
| MEASURE 1 | 0.97 | 0.95 |
| SURPASS | 0.99 | 0.97 |
Ankylosis AUC
MEASURE 10.97
SURPASS0.99
Ankylosis balanced accuracy
MEASURE 10.95
SURPASS0.97
- Bone marrow oedema (BMO) detection AUCs
- 0.85-0.93
- BMO detection balanced accuracy
- 0.72-0.74
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Automated detection of active and structural MRI lesions in the sacroiliac joints in axial spondyloarthritis: training and validation across 3 phase III clinical trial datasets. Ann Rheum Dis. Published 5 September 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →