Impact of image enhancement on grading agreement of plus disease features in retinopathy of prematurity.

Authors

Fiammetta Catania, Amandine Barjol, Mathias Gallardo, Florence Metge, Lucile Sénicourt, Giuseppe Maria Albanese, Youssef Abdel Massih, Georges Caputo, Thibaut Chapron

Published in

Graefe's archive for clinical and experimental ophthalmology = Albrecht von Graefes Archiv fur klinische und experimentelle Ophthalmologie. Jul 07, 2026. Epub Jul 07, 2026.

Abstract

To determine whether targeted image enhancement improves intergrader agreement in assessing vascular features of plus disease in retinopathy of prematurity (ROP), and to identify which specific features and clinical subgroups benefit most.
This prospective image grading study included 48 eyes from preterm infants undergoing ROP screening at a tertiary care center. All eyes were imaged with the RetCam Envision system. For each eye, five fundus images centered on the optic disc were selected and underwent standardized enhancement combining blue-channel attenuation and CLAHE-based contrast optimization. Five pediatric ophthalmologists with varying ROP experience independently evaluated raw and enhanced image sets in two grading rounds, one month apart. For each quadrant, graders recorded the presence or absence of arterial and venous tortuosity and dilation. Intergrader agreement was quantified using unweighted and weighted Fleiss' kappa and overall percentage agreement.
Image enhancement significantly improved intergrader agreement (mean weighted agreement: 61.5% for raw vs. 67.8% for enhanced images, p < 0.05). The greatest improvements were observed for arterial tortuosity (+ 11.9%) and vein tortuosity (+ 6.5%). Nasal quadrants showed the most pronounced benefit. Subgroup analyses revealed larger improvements in borderline pre-plus cases and in eyes that had previously received treatment.
Targeted image enhancement improves intergrader agreement in the evaluation of plus disease in ROP, especially for diagnostically challenging features and regions. The strongest effect is seen in borderline and treated cases, supporting the integration of standardized preprocessing techniques to enhance diagnostic objectivity for both clinicians and AI-assisted systems.

PMID:
42412199
Bibliographic data and abstract were imported from PubMed on 07 Jul 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.