Endoscopy June 2025 PMID 39884307

Impact of standard enhancement settings of endoscopy systems on performance of endoscopic artificial intelligence systems.

Martijn R Jong, Carolus H J Kusters, Querijn N E van Bokhorst, Jelmer B Jukema, Rixta A H van Eijck van Heslinga, Kiki N Fockens, Britt B S L Houwen, Tim J M Jaspers, Tim G W Boers, Manon van der Vlugt, Evelien Dekker, Fons van der Sommen, Peter H N de With, Albert J de Groof, Jacques J G H M Bergman

Full text PDF Download social card

Mentions

Peak: 2 in Jun 2025

across 2 sources since May 2025

Extended reach

23,163

potential audience

Total likes

Video views

Mentions per month

May 2025 – Jun 2025

Where it's discussed

X / Twitter2 posts
YouTube1 videos

AI synopsis

What people are saying

The discussions highlight the significance of adjusting enhancement settings to stabilize AI results, with comments referencing the methodology like retraining AI systems to handle variable settings, and terms like 'crucial' and 'potentially transformative' illustrating both the technical importance and practical implications of the findings.

AgreementModerate agreement

Most discussions acknowledge the importance of the study's findings about how equipment settings impact AI performance, showing general agreement with its conclusions.

InterestModerate interest

Participants demonstrate moderate interest, discussing technical details and implications without expressing strong enthusiasm or disinterest.

EngagementModerate engagement

Comments include references to the methodology, such as retraining AI models, indicating a meaningful level of engagement.

ImpactModerate impact

There is a shared recognition that understanding and standardizing endoscopy settings could influence clinical outcomes, reflecting a perceived moderate to high impact.

Abstract in brief

The study shows that the accuracy of AI systems used in endoscopy can significantly vary depending on the postprocessing enhancement settings of the endoscopy equipment, which may affect their real-world reliability.
Researchers tested two AI systems for detecting Barrett’s neoplasia and colorectal polyps and found that changing enhancement settings caused substantial fluctuations in their sensitivity and specificity.
By retraining the AI systems with data augmented to include different enhancement settings, they successfully reduced performance variability, suggesting a practical way to improve AI robustness in clinical practice.

Read the full abstract

All mentions

Impact of Endoscopy Settings on AI Performance in Diagnostics

Martijn Jong discusses how standard enhancement settings of endoscopy systems influence the performance of AI-based detection and diagnosis tools, highlighting the importance of optimized settings and retraining for…

endoscopythieme · May 28, 2025 74 views

Endoscopy Journal

@endoscopyjrnl · X

Jun 8, 2025 · 3 likes · 11,537 followers View post →