Advancing human-centric AI for robust X-ray analysis through holistic self-supervised learning

TL;DR

This paper introduces RayDINO, a self-supervised large visual encoder trained on 873k chest X-rays, demonstrating improved generalization, bias mitigation, and versatility across multiple radiology tasks, advancing human-centric AI in medical imaging.

Contribution

The paper presents RayDINO, a novel self-supervised model trained on a large X-ray dataset, achieving state-of-the-art performance and better robustness across diverse radiology tasks.

Findings

RayDINO outperforms previous models on nine radiology tasks.

Self-supervision enhances model generalization to unseen populations.

RayDINO reduces biases related to population, age, and sex.

Abstract

AI Foundation models are gaining traction in various applications, including medical fields like radiology. However, medical foundation models are often tested on limited tasks, leaving their generalisability and biases unexplored. We present RayDINO, a large visual encoder trained by self-supervision on 873k chest X-rays. We compare RayDINO to previous state-of-the-art models across nine radiology tasks, from classification and dense segmentation to text generation, and provide an in depth analysis of population, age and sex biases of our model. Our findings suggest that self-supervision allows patient-centric AI proving useful in clinical workflows and interpreting X-rays holistically. With RayDINO and small task-specific adapters, we reach state-of-the-art results and improve generalization to unseen populations while mitigating bias, illustrating the true promise of foundation models: versatility and robustness.