Learning Generalized Medical Image Representations through Image-Graph Contrastive Pretraining

TL;DR

This paper introduces an Image-Graph Contrastive Learning framework that leverages structured report knowledge graphs to improve medical image representations, reducing annotation needs and achieving radiologist-level performance.

Contribution

It presents a novel graph encoding strategy using relational graph convolution and transformer attention to enhance contrastive learning in medical imaging.

Findings

Outperforms existing image-text contrastive methods in linear and few-shot evaluations

Achieves performance comparable to radiologists on the CheXpert dataset

Demonstrates the effectiveness of structured clinical knowledge in medical image representation

Abstract

Medical image interpretation using deep learning has shown promise but often requires extensive expert-annotated datasets. To reduce this annotation burden, we develop an Image-Graph Contrastive Learning framework that pairs chest X-rays with structured report knowledge graphs automatically extracted from radiology notes. Our approach uniquely encodes the disconnected graph components via a relational graph convolution network and transformer attention. In experiments on the CheXpert dataset, this novel graph encoding strategy enabled the framework to outperform existing methods that use image-text contrastive learning in 1% linear evaluation and few-shot settings, while achieving comparable performance to radiologists. By exploiting unlabeled paired images and text, our framework demonstrates the potential of structured clinical insights to enhance contrastive learning for medical images. This work points toward reducing demands on medical experts for annotations, improving diagnostic precision, and advancing patient care through robust medical image understanding.