Sign in Subscribe
Artificial intelligence

Artificial Intelligence in Radiology: External Validation of Winning Algorithm Yields Promising Results

Strategic Insights

— 1 min read

A team of researchers from Ohio State University has published a study on the external validation of a winning artificial intelligence (AI) algorithm from the RSNA 2022 Cervical Spine Fracture Detection Challenge. The study, published in the American Journal of Neuroradiology, aimed to assess the feasibility of using AI algorithms in real-world clinical practice. The researchers tested a deep learning algorithm on a new dataset of 100 examinations from a level 1 trauma center and found promising results, with a sensitivity of 95%, specificity of 98%, and area under the curve of 0.97.

Key Takeaways:

  • The Radiological Society of North America has actively promoted AI challenges since 2017, with the RSNA 2022 Cervical Spine Fracture Detection Challenge producing state-of-the-art performance in its data set.
  • The research team from Ohio State University conducted a generalizability test of a leading AI algorithm from the competition, using a new dataset of 100 examinations from a level 1 trauma center.
  • The algorithm demonstrated high sensitivity (95%) and specificity (98%) in detecting cervical spine fractures, with an area under the curve (AUC) of 0.97.
  • The study highlighted the need for further research to help elucidate the potential contributions and pitfalls of AI algorithms in supporting clinical care.
  • The researchers found that the performance of the algorithm was consistent across different patient populations, with no significant differences in sensitivity and specificity.
  • The study demonstrated the feasibility of using AI algorithms in real-world clinical practice, with the algorithm processing 100 examinations in just 6.4 seconds per examination.

Statistics:

  • Sensitivity: 95%
  • Specificity: 98%
  • Area under the curve (AUC): 0.97
  • Time taken to process one examination: 6.4 seconds
  • Patient population: 50 consecutive cervical spine CT scans with at least 1 fracture present and 50 consecutive negative CT scans
  • Age range of patients: 53.5 ± 21.8 years
  • Sample size: 100 examinations
  • Trauma center source: Level 1 trauma center

Sources:

  • External Validation of a Winning Artificial Intelligence Algorithm from the RSNA 2022 Cervical Spine Fracture Detection Challenge. American Journal of Neuroradiology, 2025;46(9):1852-1858.
  • Amer Soc Neuroradiology, PO Box 3000, Denville, NJ 07834-9349, USA.

Read more