RetFiner: A Vision-Language Refinement Scheme for Retinal Foundation Models

TL;DR

RetFiner is a novel vision-language refinement scheme that enhances retinal foundation models trained on OCT images by leveraging textual data, leading to improved performance on diverse retinal disease classification tasks.

Contribution

It introduces a new SSL vision-language refinement method that boosts existing retinal foundation models' semantic understanding and downstream task performance.

Findings

Significant improvements in linear probing accuracy on OCT classification tasks.

Average performance increase of 5.8, 3.9, and 2.1 percentage points over baselines.

RetFiner enables direct adaptation of models to specific populations.

Abstract

The rise of imaging techniques such as optical coherence tomography (OCT) and advances in deep learning (DL) have enabled clinicians and researchers to streamline retinal disease staging. A popular DL approach is self-supervised learning (SSL), where models learn from vast amounts of unlabeled data, avoiding costly annotation. SSL has allowed the development of foundation models (FMs), large models that can be used for a variety of downstream tasks. However, existing FMs for OCT, trained solely on image data, lack a comprehensive and robust semantic understanding of images, as evidenced by their downstream performance (especially for complex tasks), and thus require supervised fine-tuning (which may be unfeasible) to better adapt to specific applications and populations. To address this, we propose RetFiner, an SSL vision-language refinement scheme that improves the representations of existing FMs and enables their efficient and direct adaptation to specific populations for improved downstream performance. Our method uses a diverse set of training objectives which take advantage of the rich supervisory signal found in textual data. We tested RetFiner on the retinal FMs RETFound, UrFound, and VisionFM, showing significant improvements in linear probing performance on seven highly diverse OCT classification tasks, with an average increase of 5.8, 3.9, and 2.1 percentage points over their baselines, respectively. Our code and model weights are publicly available at https://github.com/ronnief1/RetFiner.