Spondyloarthritis & psoriatic arthritis
Automated detection of spinal bone marrow oedema in axial spondyloarthritis: training and validation using two large phase 3 trial datasets
Authors' conclusion
ML software shows potential for automated MRI BMO assessment in axSpA, offering benefits such as improved consistency, reduced labour costs and minimized inter- and intra-reader variability.
Study context
- Population
- Patients with axial spondyloarthritis (axSpA) in two clinical trial datasets (radiographic and non-radiographic axSpA)
- Participants
- 686 patients3,483 spinal MRIs
- Comparison
- Machine learning (ML) models vs expert scoring of spinal MRI bone marrow oedema (BMO)
- Outcome
- Performance for presence vs absence of spinal MRI BMO (kappa, balanced accuracy, sensitivity, specificity, AUC) vs absolute reader consensus scoring
Methodological detailsmachine learning model training and validation, two clinical trial datasets
- Machine learning model training and validation
- Two clinical trial datasets
Performance
As reported in the abstract
BMO detection, r-axSpA dataset
| Test or model | AUC | Accuracy | Kappa | Inter-reader accuracy |
|---|---|---|---|---|
| Model 1 | 0.94 | 75.8% | 0.50 | — not stated in the abstract |
| Model 2 | 0.88 | 70.5% | 0.31 | — not stated in the abstract |
| Expert readers | — not stated in the abstract | — not stated in the abstract | 0.47 | 76.8% |
AUC
Model 10.94
Model 20.88
Expert readers— not stated in the abstract
Accuracy
Model 175.8%
Model 270.5%
Expert readers— not stated in the abstract
Kappa
Model 10.50
Model 20.31
Expert readers0.47
Inter-reader accuracy
Model 1— not stated in the abstract
Model 2— not stated in the abstract
Expert readers76.8%
— not stated in the abstract
Further findings
| Outcome | Model 1 | Model 2 | Expert readers |
|---|---|---|---|
| BMO detection, nr-axSpA dataset | |||
| AUC | 0.97 | 0.91 | — not stated in the abstract |
| Accuracy | 74.6% | 70% | — not stated in the abstract |
| Kappa | 0.52 | 0.27 | 0.46 |
| Inter-reader accuracy | — not stated in the abstract | — not stated in the abstract | 74.2% |
— not stated in the abstract
Original publication
Automated detection of spinal bone marrow oedema in axial spondyloarthritis: training and validation using two large phase 3 trial datasets. Rheumatology (Oxford). Published 9 June 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →