Deep learning-based identification of vertebral fracture and osteoporosis in lateral spine radiographs and DXA vertebral fracture assessment to predict incident fracture
Authors' conclusion
In summary, DL-detected pVF and osteoporosis in lateral spine radiographs and DXA VFA images enhanced fracture risk prediction in older adults.
Study context
- Population
- Older adults; 26,299 lateral spine radiographs from 9276 individuals attending a tertiary-level institution, and a DXA vertebral fracture assessment (VFA) dataset26,299 radiographs
- Participants
- 9,276 adults
- Comparison
- Deep learning (DL) detection of prevalent vertebral fracture and osteoporosis in lateral spine radiographs (VERTE-X) and DXA VFA images (KURE)
- VERTE-Xn = 9,276
- Outcome
- AUROC of DL models to detect prevalent vertebral fracture (pVF) and osteoporosis
Methodological detailsdeep learning model development and external fine-tuning, train/validation/test split
- Deep learning model development and external fine-tuning
- Train/validation/test split
Performance
As reported in the abstract
AUROC for DL detection of pVF and osteoporosis
| Test or model | AUC prevalent vertebral fracture (pVF) | AUC osteoporosis |
|---|---|---|
| VERTE-X | 0.926 (95% CI 0.908-0.955) | 0.848 (95% CI 0.827-0.869) |
| KURE | 0.924 (95% CI 0.905-0.942) | 0.867 (95% CI 0.853-0.881) |
AUC prevalent vertebral fracture (pVF)
VERTE-X0.926 (95% CI 0.908-0.955)
KURE0.924 (95% CI 0.905-0.942)
AUC osteoporosis
VERTE-X0.848 (95% CI 0.827-0.869)
KURE0.867 (95% CI 0.853-0.881)
Further findings
Incident fracture · median follow-up 5.4 years (VERTE-X) and 6.4 years (KURE)
Incident fracture risk with DL-detected vertebral fracture
| HR (95% CI) | ||
|---|---|---|
| DL-detected vertebral fracture (VERTE-X) | 3.23 (2.51–5.17) | |
| DL-detected vertebral fracture (KURE) | 2.11 (1.62–2.74) | |
| 0.20.5125Hazard ratio (log scale) |
Incident fracture risk with DL-detected osteoporosis
| HR (95% CI) | ||
|---|---|---|
| DL-detected osteoporosis (VERTE-X) | 2.62 (1.90–3.63) | |
| DL-detected osteoporosis (KURE) | 2.14 (1.72–2.66) | |
| 0.20.5125Hazard ratio (log scale) |
- Incident fracture risk with DL-detected vertebral fracture: remained significant after adjustment for clinical risk factors and femoral neck BMD
- Incident fracture risk with DL-detected osteoporosis: remained significant after adjustment for clinical risk factors and femoral neck BMD
Incident fracture discrimination
- DL scores improved incident fracture discrimination and net benefit when combined with clinical risk factors
Original publication
Deep learning-based identification of vertebral fracture and osteoporosis in lateral spine radiographs and DXA vertebral fracture assessment to predict incident fracture. J Bone Miner Res. Published 1 April 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →