# Seeing the Unseen: Artificial Intelligence-Assisted Detection of Subtle Colorectal Adenomas During Colonoscopy

## TL;DR

AI helps detect hard-to-see colorectal adenomas during colonoscopies, but raises questions about how to handle many new findings.

## Contribution

Demonstrates AI's ability to detect subtle adenomas missed in standard exams, revealing a high number of previously unseen lesions.

## Key findings

- AI-assisted colonoscopy detected 30 additional polyps not seen during conventional inspection.
- Histopathology showed mostly low-grade dysplasia tubular adenomas and one high-grade dysplasia tubulovillous adenoma.
- The case illustrates both AI's potential and challenges in managing extensive AI-detected lesions.

## Abstract

Colonoscopy is central to colorectal cancer (CRC) prevention, with adenoma detection rate (ADR) serving as a key quality indicator. Artificial intelligence (AI)-based computer-aided detection (CADe) systems have been developed to assist endoscopists by highlighting subtle mucosal abnormalities during withdrawal. We report the case of a 67-year-old male who underwent AI-assisted colonoscopy for a positive faecal occult blood test, during which CADe prompted the detection and resection of 30 additional polyps not prospectively recognised during conventional white-light inspection. Histopathological analysis demonstrated predominantly tubular adenomas with low-grade dysplasia, as well as a tubulovillous adenoma with focal high-grade dysplasia. This case highlights the potential of CADe to lower the visual threshold for adenoma detection while also illustrating the associated procedural burden and clinical uncertainties surrounding extensive AI-guided lesion detection in routine practice.