AI-assisted diagnosis of colorectal tubular adenomas using white-light, magnifying chromoendoscopy (including NBI) and pathology images

Application Evaluation Research on the Artificial Intelligence-assisted Support System for the Diagnosis of Colorectal Tubular Adenoma Lesions

This project will test whether an AI tool that combines white-light, magnified chromoendoscopy, NBI, and pathology images can help doctors identify colorectal tubular adenomas in adults undergoing colonoscopy.

Quick facts

• Study type: Observational
• Enrollment: 4000 (estimated)
• Ages: 18 Years and up
• Sex: All
• Sponsor: Renmin Hospital of Wuhan University (other)
• Locations: 1 site (Wuhan, Hubei)
• Trial ID: NCT07073430 on ClinicalTrials.gov

What this trial studies

This prospective, multi-center observational project will build a matched "trinity" database of white-light endoscopy, magnifying chromoendoscopy (including NBI), and corresponding pathological images of colorectal tubular adenomas. Investigators will apply the previously proposed multimodal endoscopic LAFEQ approach to train deep-learning diagnostic models and to create an interpretable risk-prediction model for adenomas. Models will be validated across participating hospitals to measure diagnostic accuracy and the transparency of the model's decision basis. Patients will receive standard-of-care colonoscopy and pathology; collected imaging and pathology data will be used for model development and evaluation.

Who should consider this trial

Good fit: Adults aged 18 or older who are scheduled for colonoscopy, can give informed consent, and can complete standard bowel preparation are ideal candidates.

Not a fit: Patients with prior abdominal or pelvic surgery or radiotherapy, active lower gastrointestinal bleeding, hereditary polyposis or inflammatory bowel disease, uncontrolled cardiovascular conditions, pregnancy, or inability to complete bowel preparation are unlikely to benefit or be eligible.

Why it matters

Potential benefit: If successful, the tool could help clinicians detect and characterize adenomas more accurately and provide clear, interpretable reasons for its recommendations, potentially improving diagnosis and treatment decisions.

How similar studies have performed: Prior AI research for polyp detection and characterization has shown promising accuracy, but fully multimodal and interpretable systems like this are relatively novel and not yet widely validated in large prospective multicenter cohorts.

Eligibility criteria

Inclusion Criteria:

* Patients aged ≥ 18 years, who need to undergo colonoscopy, regardless of gender.
* Voluntarily sign the informed consent form
* Promise to abide by the research procedures and cooperate in the implementation of the entire research process.

Exclusion Criteria:

* Patients who has a history of abdominal or pelvic surgery or radiotherapy in the past;
* Patients who has definite active lower gastrointestinal bleeding.
* Existing or suspected hereditary colorectal polyposis, inflammatory bowel disease;
* Uncontrolled hypertension (systolic blood pressure \> 160 mmHg or diastolic blood pressure \> 95 mmHg after standardized treatment)
* There is a history of stroke, coronary artery disease, or vascular disease;
* Pregnant;
* Intestinal preparation cannot be carried out.

Where this trial is running

Wuhan, Hubei

• Renmin Hospital of Wuhan University — Wuhan, Hubei, China (RECRUITING)

Study contacts

• Study coordinator: Mingkai Chen

Conditions: Colorectal Adenoma, Artificial Intelligence