Specialized curricula for training vision-language models in retinal image analysis

TL;DR

This paper introduces RetinaVLM, a specialized vision-language model trained with a curriculum to improve clinical decision-making in retinal image analysis, outperforming general models and approaching ophthalmologists' accuracy.

Contribution

Developed a curriculum-based training method to specialize foundation vision-language models for retinal clinical tasks, significantly enhancing their performance.

Findings

RetinaVLM outperforms general medical VLMs and ChatGPT-4o in disease staging and referral tasks.

RetinaVLM approaches the diagnostic accuracy of junior ophthalmologists.

Senior ophthalmologists found RetinaVLM's reports substantially more accurate than ChatGPT-4o.

Abstract

Clinicians spend a significant amount of time reviewing medical images and transcribing their findings regarding patient diagnosis, referral and treatment in text form. Vision-language models (VLMs), which automatically interpret images and summarize their findings as text, have enormous potential to alleviate clinical workloads and increase patient access to high-quality medical care. While foundational models have stirred considerable interest in the medical community, it is unclear whether their general capabilities translate to real-world clinical utility. In this work, we demonstrate that OpenAI's ChatGPT-4o model, in addition to two foundation VLMs designed for medical use, markedly underperform compared to practicing ophthalmologists on specialist tasks crucial to the care of patients with age-related macular degeneration (AMD). To address this, we initially identified the essential capabilities required for image-based clinical decision-making, and then developed a curriculum to selectively train VLMs in these skills. The resulting model, RetinaVLM, can be instructed to write reports that significantly outperform those written by leading foundation medical VLMs and ChatGPT-4o in disease staging (F1 score of 0.63 vs. 0.33) and patient referral (0.67 vs. 0.50), and approaches the diagnostic performance of junior ophthalmologists (who achieve 0.77 and 0.78 on the respective tasks). Furthermore, in a single-blind reader study two senior ophthalmologists with up to 32 years of experience found RetinaVLM's reports were found to be substantially more accurate than those by ChatGPT-4o (64.3% vs. 14.3%). These results reinforce that our curriculum-based approach provides a blueprint towards specializing foundation medical VLMs for real-world clinical tasks.