Effect of Two Colonoscopy AI Systems for Colon Polyp Detection

Seoul National University

Status: Completed

Conditions: Sessile Serrated Adenoma; Colonoscopy; Adenoma

Treatments: Device: Assist by artificial intelligence system for colon polyp detection

Study type: Interventional

Funder types: Other

Identifiers

NCT05089071

SNUH IRB 2107-235-1240

Details and patient eligibility

About

Computer-aided detection (CADe) systems have been actively researched for polyp detection in colonoscopy. The investigators aim to identify the effect of two CADe systems according to the system performance on false positive rate

Full description

Artificial intelligence technology based on deep learning is being applied in various medical fields, and research is being actively conducted to develop computer-aided detection (CADe) systems for colonoscopies to overcome the limitation of the variance of human skills. These well-trained CADe systems demonstrated high performance for neoplastic polyp detection and reported a 44% increase in adenoma detection rate (ADR) for endoscopists. However, the level of performance in the CADe system is not clear for expert endoscopists to be useful for ADR increase.

Furthermore, false positives(FPs) of the CADe system may negatively influence ADR during a screening colonoscopy. Accordingly, the investigators sought to identify the effect of the colonoscopy CADe system according to FP performance in endoscopists with various levels. The investigators hypothesized that the CADe system with low FPs would be useful to prevent the decrease in ADR in case of a high endoscopy workload according to the performance of CADe systems.

Enrollment: 3,046 patients

Sex: All

Ages: 45 to 100 years old

Volunteers: Accepts Healthy Volunteers

Inclusion criteria

patient for screening or surveillance colonoscopy patients agreed with participating in the study

Exclusion criteria

patients who do not agree with participating in the study patients with a history of colon resection patients with a history of inflammatory bowel resection patients with poor bowel preparation

Trial design

Primary purpose: Diagnostic

Allocation: Non-Randomized

Interventional model: Parallel Assignment

Masking: Single Blind

---

3,046 participants in 2 patient groups

CADe group

Experimental group

Description:

Endoscopists perform colonoscopy with CADe system

Treatment:

Device: Assist by artificial intelligence system for colon polyp detection

Control

No Intervention group

Description:

Endoscopists perform colonoscopy without CADe system