Related Experiment Video
Updated: May 3, 2026
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
Diagnostic performance of artificial intelligence for identification of cervical spine fractures: a systematic review
Ali Gholamrezanezhad1, Mehrdad Farrokhi2, Sami Almasri3
1Department of Radiology, Keck School of Medicine, University of Southern California (USC), Los Angeles, CA, 90089, USA.
Purpose:
This meta-analysis aimed to evaluate how effectively artificial intelligence (AI) models can diagnose cervical spine fractures.
Methods:
A systematic search from inception to July 2025 in databases of Medline, Scopus, and Web of Science was performed to retrieve studies applying AI models for diagnosis of cervical spine fractures. Deduplication, study screening, and full-text evaluations were performed using EndNote by two independent authors. Statistical analyses were conducted using MIDAS and Metaprop packages in Stata and Meta-Disc software.
Results:
Sixteen studies reporting 69 AI models were analyzed. The estimated sensitivity was 0.93 (95% CI, 0.92-0.94) and specificity was found to be 0.97 (95% CI, 0.97-0.97). The combined positive and negative likelihood ratios were 26.34 (95% CI, 15.93-43.56) and 0.07 (95% CI, 0.03-0.16), respectively. The overall accuracy was 0.96 (95% CI, 0.95-0.98), with summary receiver operating characteristic curve (AUROC) suggesting of 0.99 area under curve.
Conclusion:
Our meta-analysis showed that AI models demonstrate high diagnostic accuracy (0.96) for the early identification of cervical spine fractures. While our analysis suggests that AI models may achieve similar diagnostic performance using either CT or radiography, this finding should be interpreted with caution due to the limited number of X-ray studies and the high degree of heterogeneity. Given the scarcity of independent peer-reviewed validation studies and the variability in performance across AI models and study settings, the generalizability of these findings across different patient populations and clinical environments remains uncertain. Further research is required to confirm these results and account for study design variations.
