Research in Context

Research in Context is a structured summary that editors at The Lancet’s journals ask all authors to provide when submitting an original research article. Read more

A collaboration with

Single randomised trial: investigational AI system reduced gastric neoplasm miss rate versus standard endoscopy.

The ENDOANGEL-LD AI system reduced gastric neoplasm miss rates compared with standard upper gastrointestinal endoscopy; the authors describe potential future support for targeted biopsies.

  • Lianlian Wu, MD; Renduo Shang, MD; Prateek Sharma, MD; Wei Zhou, MD; Jun Liu, MM; Liwen Yao, MD; et al.

  • July 20, 2021

Share

Evidence before this study

We searched PubMed for papers published between Jan 1, 2000, and Jul 1, 2020, with the keywords “artificial intelligence” OR “deep learning” AND “endoscopy”. No restrictions on study type or language were implemented. In recent years, great progress has been made in lesion detection during endoscopies assisted by deep learning algorithms. This is primarily based on research on artificial intelligence (AI) in colorectal polyp detection and classification from a series of clinical trials designed to verify the effectiveness of AI in assisting endoscopists during colonoscopy. However, for the application of AI in upper gastrointestinal endoscopy, most reported studies are still in the early preclinical stage. Several preliminary studies of the use of deep learning for diagnosing early-stage gastric cancer have reported promising results but need to be confirmed by prospective clinical trials, as most algorithms were retrospectively tested in still images.

Added value of this study

Here, we report the results of a randomised controlled tandem trial to evaluate the effectiveness of the ENDOANGEL-LD AI system for the detection of focal lesions and predicting neoplasms in the stomach using white light endoscopy. Compared with standard upper gastrointestinal endoscopy, the AI system significantly reduced the miss rate of gastric neoplasms (relative risk 0·224, 95% CI 0·068–0·744). To the best of our knowledge, this is the first study to evaluate the effect of a deep learning system on improving neoplasm detection by upper gastrointestinal endoscopy in a randomised controlled trial.

Implications of all the available evidence

AI systems such as ENDOANGEL-LD could effectively reduce the miss rate of gastric neoplasia and minimise unnecessary biopsies without increasing procedure time. Our system has the potential to assist endoscopists in obtaining targeted biopsies and reducing gastric neoplasm miss rates in clinical practice.

Source: The Lancet Gastroenterology & Hepatology

Your research deserves a global audience

The Lancet Group reaches millions of clinicians and researchers worldwide.

Publish with The Lancet Group

Related Content

Regorafenib combination explored in untreated metastatic oesophagogastric cancer

An early, first-in-setting study suggests adding investigational regorafenib to nivolumab and chemotherapy may be safe and active; a randomised phase 3 trial is planned to confirm these preliminary findings.

  • by Samuel L Cytryn, MD; Ryan H Moy, MD; Darren Cowzer, MBBCh; Ronak H Shah, MS; Joanne F Chou, MPH; Smita S Joshi, MD; et al.

  • September 1, 2023

Gloved hand holding a blood sample tube.
Phase 3 SPOTLIGHT trial: investigational zolbetuximab shows survival benefit in CLDN18.2-positive gastric cancer

In patients with CLDN18.2-positive, HER2-negative advanced gastric or gastro-oesophageal junction adenocarcinoma, adding investigational zolbetuximab to chemotherapy prolonged progression-free and overall survival versus chemotherapy alone, with more nausea and vomiting leading to discontinuation.

  • by Kohei Shitara, MD; Florian Lordick, MD; Yung-Jue Bang, MD; Peter Enzinger, MD; David Ilson, MD; Manish A Shah, MD; et al.

  • April 14, 2023

Illustration highlighting the stomach in the human body.