Disclosures: The authors report no relevant financial disclosures.
August 10, 2021
1 min read
Save

AI-assisted endoscopy reduces gastric neoplasm miss rate

Disclosures: The authors report no relevant financial disclosures.
You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Artificial intelligence-assisted upper gastrointestinal endoscopy reduced the miss rate for gastric neoplasms, according to research published in Lancet Gastroenterology and Hepatology.

“Achieving a low miss rate and a low biopsy rate at the same time seems to be an unrealistic goal in traditional endoscopy. ... AI has shown its potential in endoscopic diagnosis,” Lianlian Wu, MD, Renmin Hospital of Wuhan University, and colleagues wrote. “In a previous study, our team developed an AI system named ENDOANGEL with a gastric cancer detection module that was evaluated in a preliminary multicenter clinical trial. Thereafter, we further modified and improved the model to identify gastric neoplasms and the updated system was renamed ENDOANGEL-LD (lesion detection).”

Gastric neoplasm miss rate

In a randomized-controlled, tandem trial, researcher evaluated the effect of ENDOANGEL-LD on improved neoplasm detection among 1,812 patients undergoing routine upper GI endoscopy. Patients underwent either AI-assisted white light endoscopy (n = 907) or routine white light endoscopy (n = 905) followed immediately by the other procedure; targeted biopsies for all detected lesions followed the second examination. The primary endpoint was the gastric neoplasm miss rate.

According to study results, the combined number of diagnoses was 49 gastric neoplasms in 47 patients in the AI group and 44 gastric neoplasms in 43 patients in the routine group. Researchers noted a lower miss rate among the AI group vs. the routine group with 6.4% (95% CI, 1.6-17.9) of patients with neoplasms missed by AI-assisted endoscopy first vs. 25.6% (95% CI, 15.5-43) of patients with neoplasms missed by routine endoscopy first (RR = 0.224; 95% CI, 0.068-0.744). The only adverse event was bleeding from a targeted lesion following biopsy.

“ENDOANGEL-LD, an AI system that detects focal lesions and predicts gastric neoplasms, could effectively reduce the miss rate of gastric neoplasms and minimize unnecessary biopsies without adding inspection time. It has the potential to assist endoscopists in lesion detection and reducing the gastric neoplasm miss rate in clinical practice,” Wu and colleagues concluded. “Further studies with a larger sample size and more centers should be done to determine the scalability and generalizability of this system in reducing the miss rate of gastric neoplasms.”