AI-Assisted Real-Time Endoscopic Characterization of Diminutive Colorectal Polyps in Non-Academic Hospitals

Shanghai Jiao Tong University School of Medicine

Status: Begins enrollment this month

Conditions: Colorectal Polyps

Treatments: Device: AI-Assisted Real-time Endoscopic Diagnosis System

Study type: Interventional

Funder types: Other

Identifiers

NCT07362524

KY2024-080-A

Details and patient eligibility

About

To validate the feasibility of promoting colorectal diminutive polyp "predict, resect, and discard" and "diagnose and leave" strategies for non-experts in general hospitals through CADx system.

Full description

This study will: (1) develop and refine a deep learning model for optical diagnosis of diminutive colorectal polyps-predicting histologic type and the presence of advanced adenoma components-based on multicenter endoscopic videos; (2) conduct a randomized controlled study recruiting endoscopists from different regions and experience levels nationwide to evaluate, with or without access to the model's diagnostic output, the success rates of implementing the "resect and discard" and "diagnose and leave" strategies as well as identifying advanced adenoma components; and (3) carry out a prospective randomized controlled trial to assess the benefits of a real-time AI-assisted endoscopic system in facilitating endoscopists' implementation of the "resect and discard" and "diagnose and leave" strategies for diminutive colorectal polyps.

Enrollment: 363 estimated patients

Sex: All

Ages: 18+ years old

Volunteers: No Healthy Volunteers

Inclusion criteria

1. Agree to participate in this prospective randomized controlled study;
2. Age ≥18 years for colonoscopy;
3. At least one colorectal diminutive polyp (size <=5mm) found during examination;

Exclusion criteria

1. Already participating in other clinical trials, having signed informed consent and being in the follow-up period of other clinical trials;
2. Already participating in drug clinical trials and being in the washout period of experimental or control drugs;
3. History of drug or alcohol abuse or psychological disorders in the past five years;
4. Pregnant or lactating patients;
5. Known polyposis syndromes;
6. Patients with gastrointestinal bleeding;
7. Previous history of inflammatory bowel disease, colorectal cancer, or colorectal surgery;
8. Patients with contraindications to tissue biopsy;
9. Previous history of allergic reactions to bowel cleansing agent components;
10. Suffering from intestinal obstruction or perforation, toxic megacolon, heart failure (Grade III or IV), severe cardiovascular disease, severe liver failure, or renal insufficiency, etc.;
11. Patient's bowel preparation BBPS score <6 for this colonoscopy;
12. Researchers consider the patient unsuitable for trial participation.

Trial design

Primary purpose: Diagnostic

Allocation: Randomized

Interventional model: Parallel Assignment

Masking: Single Blind

---

363 participants in 2 patient groups

CADx assisted colonoscopy group

Experimental group

Treatment:

Device: AI-Assisted Real-time Endoscopic Diagnosis System

Routine colonoscopy group

No Intervention group

Description:

Endoscopists perform standard endoscopic diagnosis using conventional NBI-based NICE classification without AI assistance.