Deep Learning for Opportunistic Vertebral Fracture Detection on Routine Thoraco-abdominal Computed Tomography: A Systematic Review and Hierarchical Summary Receiver Operating Characteristic Meta-analysis of Patient-level Diagnostic Test Accuracy.
Kalab Yigermal Gete, Zulaika Nassir Idris, Belete Achamyelew Ayele
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis evaluated the diagnostic accuracy of deep learning (DL) for detecting vertebral fractures (VFs) on routine thoraco-abdominal CT in a population of 11,615 patients across seven studies. The pooled sensitivity was 0.83 and specificity was 0.92, indicating that DL can effectively identify VFs, although the evidence is still early and heterogeneous.
Key findings
- Pooled sensitivity for VF detection was 0.83 (95% CI: 0.73-0.90).
- Pooled specificity for VF detection was 0.92 (95% CI: 0.90-0.94).
- The positive likelihood ratio was 10.44, indicating strong diagnostic performance.
- Risk of bias was assessed as low to moderate across included studies.
Why it matters
The ability to accurately detect vertebral fractures using DL on routine CT could enhance clinical decision-making and improve patient outcomes by facilitating timely interventions for fractures that are often overlooked.
What to keep in mind
The evidence is based on small, retrospective studies with heterogeneous populations and incomplete AI-specific reporting.