Artificial intelligence and upper gastrointestinal endoscopy: Current status and future perspective

Area of Science:

• Gastroenterology
• Medical Imaging
• Artificial Intelligence

Background:

• Current gastroscopy has limitations, including a 10% false-negative rate for gastric cancer detection.
• Accurate diagnosis of Barrett's esophagus-related dysplasia requires specialized endoscopic imaging expertise.
• Artificial intelligence (AI) and computer-aided diagnosis (CAD) are emerging technologies in gastrointestinal endoscopy.

Purpose of the Study:

• To review current literature on CAD for upper gastrointestinal endoscopy.
• To summarize research from an endoscopist's perspective.
• To identify future research needs for clinical implementation.

Main Methods:

• Systematic review of publications on CAD in upper gastrointestinal endoscopy.
• Analysis of research focusing on dysplasia in Barrett's esophagus and early gastric cancer detection.
• Evaluation of studies from both engineering and medical perspectives.

Main Results:

• CAD technologies aim to assist endoscopists in identifying neoplastic changes in the esophagus and stomach.
• Existing studies, though often ex vivo, show promising results.
• Real-time, on-site decision support for endoscopists is the ideal application for CAD.

Conclusions:

• CAD holds potential to enhance diagnostic accuracy and overcome skill-based limitations in gastroscopy.
• The clinical feasibility, effectiveness, and safety of CAD require further investigation.
• Future research should focus on bridging the gap between experimental findings and practical clinical application.