Saved in:
Bibliographic Details
Main Author: Cho, Sungjun
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.13403
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915861504196608
author Cho, Sungjun
author_facet Cho, Sungjun
contents Diabetic retinopathy (DR) is a leading cause of preventable blindness, and automated fundus image grading can play an important role in large-scale screening. In this work, we investigate three CLIP-based approaches for five-class DR severity grading: (1) a zero-shot baseline using prompt engineering, (2) a hybrid FCN-CLIP model augmented with CBAM attention, and (3) a ranking-aware prompting model that encodes the ordinal structure of DR progression. We train and evaluate on a combined dataset of APTOS 2019 and Messidor-2 (n=5,406), addressing class imbalance through resampling and class-specific optimal thresholding. Our experiments show that the ranking-aware model achieves the highest overall accuracy (93.42%, AUROC 0.9845) and strong recall on clinically critical severe cases, while the hybrid FCN-CLIP model (92.49%, AUROC 0.99) excels at detecting proliferative DR. Both substantially outperform the zero-shot baseline (55.17%, AUROC 0.75). We analyze the complementary strengths of each approach and discuss their practical implications for screening contexts.
format Preprint
id arxiv_https___arxiv_org_abs_2603_13403
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Diabetic Retinopathy Grading with CLIP-based Ranking-Aware Adaptation:A Comparative Study on Fundus Image
Cho, Sungjun
Computer Vision and Pattern Recognition
Diabetic retinopathy (DR) is a leading cause of preventable blindness, and automated fundus image grading can play an important role in large-scale screening. In this work, we investigate three CLIP-based approaches for five-class DR severity grading: (1) a zero-shot baseline using prompt engineering, (2) a hybrid FCN-CLIP model augmented with CBAM attention, and (3) a ranking-aware prompting model that encodes the ordinal structure of DR progression. We train and evaluate on a combined dataset of APTOS 2019 and Messidor-2 (n=5,406), addressing class imbalance through resampling and class-specific optimal thresholding. Our experiments show that the ranking-aware model achieves the highest overall accuracy (93.42%, AUROC 0.9845) and strong recall on clinically critical severe cases, while the hybrid FCN-CLIP model (92.49%, AUROC 0.99) excels at detecting proliferative DR. Both substantially outperform the zero-shot baseline (55.17%, AUROC 0.75). We analyze the complementary strengths of each approach and discuss their practical implications for screening contexts.
title Diabetic Retinopathy Grading with CLIP-based Ranking-Aware Adaptation:A Comparative Study on Fundus Image
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.13403