Sign in Subscribe
Artificial intelligence

AI Exceeds Human Performance in Evaluating Crohn's Disease Endoscopies

Strategic Insights

— 1 min read

A study published in Clinical Gastroenterology and Hepatology has found that artificial intelligence can match and potentially exceed the performance of gastroenterologists in evaluating endoscopies of Crohn's disease patients. According to the research, a computer vision model identified mucosal ulceration as accurately as physicians reviewing videos, while also being strongly correlated with the most common endoscopic scoring system. This breakthrough could provide better solutions for more precise measurements and descriptions of what is being seen on the screen during colonoscopy.

Key Takeaways:

  • A computer vision model identified mucosal ulceration as accurately as gastroenterologists reviewing videos.
  • The AI model's results had a higher overall DICE similarity score with ground truth (0.591) than when comparing the agreement between two gastroenterologists (0.462).
  • The computer vision assessments of the colonoscopy videos were strongly correlated with Simple Endoscopic Score for Crohn's Disease (SES-CD) values.
  • The study used 4,487 still images from endoscopic videos, which were reviewed by two gastroenterologists and the AI model.
  • The AI model's ability to match human performance in evaluating endoscopies could lead to more quantitative detail to describe disease.
  • Researchers hope that AI-powered metrics will improve consistency between providers and provide more objective evidence for an experienced clinician's impression.
  • The study's findings have implications for areas without IBD specialists, where AI endoscopy interpretation may be used to guide treatment change based on expert decision making.

Statistics:

  • 0.591: The overall DICE similarity score of the AI model with ground truth.
  • 0.462: The overall DICE similarity score of the agreement between two gastroenterologists.
  • 4,487: The number of still images from endoscopic videos reviewed by two gastroenterologists and the AI model.
  • 2025: The year of publication for the study.

Sources:

  • Clinical Gastroenterology and Hepatology: "Artificial intelligence can match and potentially exceed the performance of gastroenterologists in evaluating endoscopies of Crohn's disease patients."
  • Ryan W. Stidham, M.D., M.S., associate professor of internal medicine at the University of Michigan Medical School: "A clinician has a gestalt of whether someone's Crohn's is mild or severe, if it's improving or worsening, But the manual tools to quantify that expert opinion with objectivity and reproducibly are poor. AI image analysis can provide better solutions for more precise measurements and descriptions of what we are seeing on the screen during colonoscopy."

Read more