Evaluation of the impact of artificial intelligence-assisted image interpretation on the diagnostic performance of clinicians in identifying endotracheal tube position on plain chest X-ray: a multi-case multi-reader study
Authors' conclusion
Use of assistive AI technology improved accuracy and confidence in interpreting ETT placement on CXR, especially for identification of critically misplaced tubes.
Study context
- Population
- Chest X-Ray (CXRs) of intubated adults from John Radcliffe Hospital (Oxford) and two other UK NHS hospitals, read by 18 clinical readers of varying seniority from six specialties
- Participants
- 400 CXRs
- Comparison
- Artificial intelligence (AI)-assisted vs unassisted clinician interpretation of ETT position on CXR
- Outcome
- Reader accuracy for ETT placement classification (correct, too low, too high) vs thoracic radiologist consensus
Methodological detailsmulti-case multi-reader, retrospective image sourcing, four-week washout
- Multi-case multi-reader
- Retrospective image sourcing
- Four-week washout
Performance
As reported in the abstract
Pooled accuracy for tube placement classification
- Pooled accuracy improved from 73.6 to 77.4%
- AI assistance vs Blinded classification: P = 0.002
- 14,400 image interpretations
Further findings
| Outcome | P value |
|---|---|
| Accuracy for identification of critically misplaced tubesAI assistance vs Blinded classification | P = 0.001 |
Accuracy for identification of critically misplaced tubes
- Accuracy increased from 79.3 to 89.0%
Reader confidence and interpretation time
- No change in mean interpretation time per image
- 36 s
- Reader confidence improved with AI assistance
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Evaluation of the impact of artificial intelligence-assisted image interpretation on the diagnostic performance of clinicians in identifying endotracheal tube position on plain chest X-ray: a multi-case multi-reader study. Crit Care. Published 28 July 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →