Related Experiment Video
Updated: Oct 9, 2025
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
Improving Radiographic Fracture Recognition Performance and Efficiency Using Artificial Intelligence
Ali Guermazi1, Chadi Tannoury1, Andrew J Kompel1
1From the Departments of Radiology (A. Guermazi, A.J.K., A.M.M., H.R., A.C.M., D.H.), Orthopaedic Surgery (C.T., X.L.), and Family Medicine (D.C.), Boston University School of Medicine, Boston, Mass; Department of Radiology, VA Boston Healthcare System, 1400 VFW Parkway, Suite 1B105, West Roxbury, MA 02132 (A. Guermazi); Gleamer, Paris, France (A.D., A.T., E.L., A.P., N.E.); Department of Biostatistics, CHU Rouen, Rouen, France (A. Gillibert); Department of Rheumatology, Harvard Vanguard Medical Associates, Braintree, Mass (Y.L.); Department of Radiology, Musculoskeletal Division, Massachusetts General Hospital, Harvard Medical School, Boston, Mass 02114 (M.J.); Sorbonne Université, CNRS, Institut des Systèmes Intelligents et de Robotique, Paris, France (A.P.); Department of Orthopaedic Surgery, The Mount Sinai Hospital, New York, NY (R.L.P.); University Health Services and Primary Care Sports Medicine, Boston College, Chestnut Hill, Mass (D.C.); and Department of Radiology, Stony Brook University Renaissance School of Medicine, Stony Brook, NY (D.H.).
Artificial intelligence (AI) significantly improved fracture detection sensitivity and specificity in radiographs for physicians. This AI assistance enhanced diagnostic performance without increasing reading time, aiding both radiologists and nonradiologists.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence in Medicine
Background:
- Missed fractures are a frequent cause of diagnostic discrepancies in radiographic interpretations.
- This study addresses the diagnostic performance gap in fracture detection on radiographs.
Purpose of the Study:
- To evaluate the impact of artificial intelligence (AI) assistance on physician diagnostic performance for detecting fractures on radiographs.
- To assess improvements in sensitivity and specificity with AI support.
Main Methods:
- Retrospective diagnostic study utilizing a multi-reader, multi-case methodology with a multicenter dataset of 480 examinations.
- Inclusion of diverse body regions: foot/ankle, knee/leg, hip/pelvis, hand/wrist, elbow/arm, shoulder/clavicle, rib cage, and spine.
- Twenty-four physicians interpreted cases with and without AI assistance, with a 1-month washout period.
Main Results:
- AI assistance significantly increased sensitivity per patient by 10.4% (75.2% with AI vs. 64.8% without AI; P < .001).
- Specificity per patient showed noninferiority with AI aid (+5.0%; P = .001), reaching 95.6% compared to 90.6% without AI.
- AI reduced average reading time by 6.3 seconds per examination (P = .046).
Conclusions:
- AI assistance demonstrably enhances the sensitivity of fracture detection for physicians interpreting radiographs.
- AI support also shows potential for improving specificity without prolonging interpretation time.
- The findings suggest AI is a valuable tool for improving diagnostic accuracy in fracture detection.
