Investigating Artificial Intelligence detection of unreported radiographic markers of osteoporosis on chest X-ray
Author
Luchs, Jonathan Stephen | Premier Radiology Services, US
Scientific presentation (R1-SSCH09-5) at RSNA 2023, 26 – 30. November 2023 in Chicago, US
Purpose
Can an off-label use of an AI-tool for chest X-rays detect unreported cases of osteoporosis?
Method
Retrospective analysis of 1,223 CXR cases of patients ≥65 years from US outpatient centers. Comparison of osteopenia and spine wedge fracture detected by Annalise Enterprise CXR and those findings reported in the clinical report. A ground-truth review was conducted in cases of disagreement between the model and report.
Results
Model and report agreed in 84.5% and 88.4% of cases for osteopenia and spinal wedge fracture, respectively. Of a total of 102 unreported cases, the AI model detected 95.1% and 100%, respectively.
Conclusion
AI may improve patient outcomes by flagging findings omitted by radiologists’ reports for further evaluation.
Disclaimer
Harrison.ai Radiology Solutions were previously marketed as Annalise.ai solutions.
