You're viewing our updated article page. If you need more time to adjust, you can return to the old layout.

ORIGINAL RESEARCH article

Front. Med.

Sec. Ophthalmology

Few-Shot Annotation Correction for Lightweight Retinal Vessel Image Segmentation

  • 1. Harbin Medical University, Harbin, China

  • 2. Durham University, Durham, United Kingdom

Article metrics

View details

124

Views

The final, formatted version of the article will be published soon.

Abstract

ABSTRACT Retinal vessel segmentation underpins quantitative analysis in ophthalmology and is widely used for screening and diagnosis. In practice, manual annotations for thin and tortuous vessels are error prone, yet the effect of positional label noise on segmentation quality remains underexplored. We address this gap with a lightweight few shot U Net based framework for annotation correction and noise robust learning. Analyses on DRIVE reveal clear performance degradation as label displacement increases. Cross dataset validation shows that the proposed method attains Accuracy of 96.51, AUC of 98.01, F1 of 83.55 on CHASE DB1 and Accuracy of 97.54, AUC of 98.45, F1 of 83.11 on STARE, achieving competitive performance against state of the art methods. These results quantify the sensitivity of vessel segmentation to positional annotation errors and demonstrate practical robustness under noisy labels.

Summary

Keywords

deep learning, Medical Image Analysis, Noisy annotations, Retinal vessel segmentation, U-net

Received

09 August 2025

Accepted

29 January 2026

Copyright

© 2026 Li, Sun, Organisciak, Long, Su and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Ying Su

Disclaimer

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Outline

Share article

Article metrics