Researchers found that polyp detection was significantly improved when less experienced physicians performed AI-assisted colonoscopies. Using AI in this diagnostic tool may reduce the likelihood of missing these potential precursors of colorectal cancer.
The addition of AI has enhanced several medical diagnostic tools such as mammography, ultrasound, and MRI. Now it’s colonoscopy’s turn to receive a computer-assisted upgrade.
A colonoscopy (a test in which an endoscope is inserted into the colon to examine the lining of the colon) can reduce colorectal cancer-related deaths by detecting and removing precancerous polyps, also known as adenomas. reduce the incidence of However, colonoscopy may be an imperfect diagnostic tool. Up to 26% of adenomas and 9% of advanced adenomas may be missed, increasing the risk of adverse outcomes and death. Reasons for missed adenomas include flat morphology, inadequate bowel preparation, and insufficient experience of the endoscopist.
Researchers at the Chinese University of Hong Kong (CUHK) School of Medicine now investigate whether AI-assisted colonoscopies can improve adenoma detection rate (ADR) when performed by less experienced physicians. I investigated whether.
“Our study is important for the future development of AI in clinical medicine and endoscopic training,” said Louis Lau Hosin, lead author of the study. “Junior endoscopists are generally less skilled and require higher levels of assistance during the early learning stages. Our research shows that the use of AI allows them to practice their skills in a more standardized way. This is important for endoscopic training as it has been shown to provide image guidance for It has been confirmed that it is useful.”
The AI used was computer-assisted polyp detection (CADe). This is a deep learning system that has been reported in previous trials to provide significant clinical benefits for real-time adenoma detection. From April 2021 to July 2022, researchers hired 22 young endoscopists with less than 500 endoscopy experiences and less than 3 years of training to develop an AI-assisted endoscopy system. We studied the performance using . Endoscopists were stratified into novice (<200 procedures) and intermediate (200–500 procedures) groups.
The primary endpoint of this study was ADR. Secondary endpoints included ADR for adenomas of various sizes (<5 mm, 5–10 mm, >10 mm) and locations. Advanced adenomas were precursors to colorectal cancer and were defined as 10 mm or larger.
Endoscopists in training performed colonoscopies on 766 patients. 386 were assigned to the CADe group, and the remainder underwent conventional colonoscopy. Overall, ADR was significantly higher in the CADe group compared to the control group, 57.5% vs. 44.5%, respectively. His ADR for adenomas smaller than 5 mm was 40.4% in the CADe group versus 25.0% in the control group. For adenomas between 5 and 10 mm, the rates were 36.8% and 29.2%, respectively. There was no significant difference in ADR for advanced adenomas. ADR was higher in novice (60.0% vs. 41.9%) and intermediate-level endoscopists (56.5% vs. 45.5%) in the CADe group.
Researchers say the benefit of CADe for highly advanced adenomas is still unclear. They optimize the performance of the algorithm and recommend the simultaneous development of a computer-assisted adenoma diagnosis system. Nevertheless, based on their findings, they argue for the incorporation of AI devices into endoscopy training curricula.
The study was published in the journal Clinical gastroenterology and hepatology.
sauce: Hong Kong