MedPath

Ability of Physicians to Distinguish Real From Artificial Colon Polyp Images

Not Applicable
Completed
Conditions
Colon Polyp
Colonic Adenoma
Registration Number
NCT07108569
Lead Sponsor
Wuerzburg University Hospital
Brief Summary

Training in endoscopy is essential for the early detection of precursors of colorectal cancer. Up to now, this training has been carried out with image collections of findings and in practice when working on patients.

The investigators want to use artificial intelligence (AI) to better train doctors to recognise these precursors. By using generative AI, the investigators were able to create realistic images that comply with data protection regulations and whose content can be predefined. Parts of the image can also be regenerated so that it is possible to create different precancerous stages in the same place in the image.

In this study the investigators want to identify the ability of physicians to distinguish artificial from real polyp images.

Detailed Description

Not available

Recruitment & Eligibility

Status
COMPLETED
Sex
All
Target Recruitment
53
Inclusion Criteria
  • Physicians with or without experience in colonoscopy
Exclusion Criteria

Not provided

Study & Design

Study Type
INTERVENTIONAL
Study Design
SINGLE_GROUP
Primary Outcome Measures
NameTimeMethod
Ability to detect artificial images as artificial6 months

The ability to recognise artificial images as being artificial, using an online questionnaire - binary question

Secondary Outcome Measures
NameTimeMethod
Accuracy to correctly classify images6 months

Accuracy to correctly classify images using an online questionnaire

Ability to detect real images as real6 months

The ability to recognise real images as being real using an online questionnaire - binary question

Trial Locations

Locations (1)

University hospital Würzburg b

🇩🇪

Würzburg, Germany

University hospital Würzburg b
🇩🇪Würzburg, Germany

MedPath

Empowering clinical research with data-driven insights and AI-powered tools.

© 2025 MedPath, Inc. All rights reserved.