Related Experiment Video
Updated: Jun 17, 2026
05:49
Reliability of Artificial Intelligence-Based Cone Beam Computed Tomography Integration with Digital Dental Images
Published on: February 23, 2024
The AI implementation gap in trauma radiography: standalone versus discretionary AI-integrated fracture detection
Anna Gurabi1, Sirus S Hosseini-Begtary1,2, Peter Hegedus1
1Department of Radiology, Clinic for Medical Imaging, Semmelweis University, Budapest, Hungary.
European Radiology Experimental
|June 16, 2026
Summary
Artificial intelligence (AI) shows near-expert accuracy in detecting fractures on trauma radiographs. However, its discretionary use in clinical practice does not consistently improve radiologist sensitivity, highlighting implementation challenges.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence in Healthcare
Background:
- Artificial intelligence (AI) offers potential for improving fracture detection in emergency trauma care.
- AI may assist radiologists, potentially reducing workload and enhancing diagnostic accuracy.
Purpose of the Study:
- To evaluate the role and impact of deep learning-based AI decision-support software in reporting trauma radiography cases.
- To assess the performance of AI in fracture detection compared to radiologists in a real-world setting.
Main Methods:
- Retrospective analysis of 2317 trauma radiographs from a single center.
- Comparison of AI performance with and without radiologist input, against a ground truth established by an experienced musculoskeletal radiologist.
- Calculation of accuracy, sensitivity, specificity, PPV, and NPV at both fracture and patient levels.
Main Results:
- Standalone AI achieved high accuracy (up to 94.1%) and sensitivity (up to 93.5%) in fracture detection.
- AI-assisted radiologists showed marginal improvements over non-AI-assisted radiologists, but sensitivity was not consistently enhanced.
- Discrepancies between AI and radiologists were noted, often due to anatomical variants.
Conclusions:
- Standalone AI demonstrates near-expert performance in trauma fracture detection.
- Discretionary AI integration does not guarantee improved radiologist sensitivity, indicating a gap between algorithmic performance and clinical benefit.
- Effective real-world implementation strategies are crucial for realizing the full clinical benefit of AI in radiography.