A Text-Guided Lesion Mining Vision–Language Ordinal Classification Framework for Diabetic Retinopathy Grading
Jiawen Huang, Jinxia Shang · 2026
Read A Text-Guided Lesion Mining Vision–Language Ordinal Classification Framework for Diabetic Retinopathy Grading by Jiawen Huang, Jinxia Shang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13608726.txt.
Description
Jiawen Huang and Jinxia Shang present a 2026 framework for grading diabetic retinopathy, an eye disease that can blind if caught late. Automated grading supports large screening programs, but most systems read only the image. They also ignore that severity grades form an ordered ladder, and that one grade can look quite different from patient to patient.
Their system pairs images with text. Prompts describing healthy tissue and lesion types steer the model toward the spots that matter for grading. A special branch enforces the grade order, and a multi-center constraint captures the varied looks inside one grade. On a public benchmark the method beat its base model on accuracy and related scores. Across six data sets under a leave-one-out protocol, it posted the best average accuracy reported so far.
Catalog record
| CREATOR | Jiawen Huang, Jinxia Shang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |