Clinical Evaluation of Computer-Aided Colorectal Neoplasia Detection Using a Novel Endoscopic Artificial...

Area of Science:

• Gastroenterology
• Medical Imaging
• Artificial Intelligence

Background:

• Computer-aided diagnostic systems are increasingly utilized in gastrointestinal endoscopy.
• This study evaluates a novel artificial intelligence system for computer-aided detection (CADe) of colonic adenomas.

Purpose of the Study:

• To assess the clinical performance of a new CADe system in detecting colonic adenomas.
• To compare the adenoma detection rate (ADR) and adenoma miss rate for the rectosigmoid colon (AMRrs) between the CADe and control groups.

Main Methods:

• A single-center prospective randomized study involving 415 participants.
• Participants were allocated to either the CADe group (n=207) or the control group (n=208).
• Endoscopic examinations were performed by experienced endoscopists, with performance assessed by ADR and AMRrs.

Main Results:

• The ADR was significantly higher in the CADe group (59.4%) compared to the control group (47.6%) (p=0.018).
• The AMRrs was lower in the CADe group (11.9%) versus the control group (26.0%) (p=0.037).
• No increase in examination time or need for additional staff was observed with the CADe system.

Conclusions:

• The CADe system improved ADR by 11.8% compared to experienced endoscopists alone.
• The novel CADe system offers a promising advancement for colorectal cancer diagnosis.
• This AI system enhances diagnostic effectiveness without prolonging procedures or requiring extra personnel.