Collecting colonoscopy images of large colorectal polyps for AI development
Multicenter Prospectively Collected Registry of Images and Videos of Colonic Lesions to Develop Artificial Intelligence to Predict Submucosal Invasion
This project will try to collect colonoscopy images and videos of large colorectal polyps from adults undergoing colonoscopy to help build an AI that predicts polyp histology and depth of invasion.
Quick facts
| Study type | Observational |
|---|---|
| Enrollment | 5000 (estimated) |
| Ages | 18 Years and up |
| Sex | All |
| Sponsor | Universitaire Ziekenhuizen KU Leuven Academic / other |
| Locations | 1 site (Leuven) |
| Trial ID | NCT07401264 on ClinicalTrials.gov |
What this trial studies
This is a prospective, multicenter image registry that captures digital photos and videos of colorectal polyps during routine or therapeutic colonoscopies. Sites will submit high-quality white-light or enhanced imaging of polyps at least 10 mm in size or those suspected of malignancy, together with the corresponding histology reports. The image and histology pairs will be used to train and validate artificial intelligence models to predict microscopic features and submucosal invasion. The project is led by UZ Leuven and is part of the broader ECOPOP initiative, with additional European and non-European centers planned to contribute.
Who should consider this trial
Good fit: Adults scheduled for a standard or therapeutic colonoscopy that produces digital images or videos showing at least one colorectal polyp 10 mm or larger or with suspected malignancy/submucosal invasion are appropriate candidates.
Not a fit: Patients who only have polyps smaller than 10 mm, cannot undergo colonoscopy, or whose images/videos are unusable due to poor quality are unlikely to contribute useful data or benefit directly from this project.
Why it matters
Potential benefit: If successful, the AI could help clinicians noninvasively predict whether a polyp is benign or invasive, supporting faster and more targeted treatment decisions.
How similar studies have performed: Previous AI projects using curated colonoscopy images have shown promising results in predicting polyp histology and invasion, but large prospective multicenter image registries remain relatively new.
Eligibility criteria
Show full inclusion / exclusion criteria
Inclusion Criteria: * Referral for standard or therapeutic colonoscopy * Digital video material of standard colonoscopy containing at least one colorectal polyp (≥ 10mm) and/or with suspicion of malignancy/submucosal invasion * Digital images of colorectal polyps ≥ 10mm made during standard colonoscopy and/or with suspicion of malignancy/submucosal invasion Page 11 of 21 * Videos and/or images can be made in white light or any virtual of dye-based enhancement technique * Colonoscopies performed after IRB approval for this particular study. * Material collected in adult patients of all sex or race, including pregnant women. Exclusion Criteria: * Any contraindication to undergo a standard colonoscopy * Any uncontrolled coagulopathy or bleeding disorder * Colonoscopy videos not containing any or colorectal polyps \< 10mm * Colonoscopy videos or images of low quality due to unstable imaging, stool remnants, bowel or patient movements or blurry vision * Colonoscopy videos with visualization of the polyp less than 5 seconds due to unstable positioning, passing stool remnants, bowel or patient movements or blurry visions
Where this trial is running
Leuven
- UZ Leuven — Leuven, Belgium (Recruiting)
Study contacts
- Study coordinator: Raf Bisschops, MD
- Email: raf.bisschops@uzleuven.be
- Phone: +32 16 342161
How to participate
- Review the eligibility criteria above with your treating physician.
- Visit the official trial page on ClinicalTrials.gov for the most current contact information and recruitment status.
- Contact the listed study coordinator or principal investigator to request pre-screening. Pre-screening is free and never obligates you to enroll.