Disentangling representations of retinal images with generative models

TL;DR

This paper introduces a generative model that disentangles patient attributes from camera effects in retinal fundus images, improving controllability and realism for ophthalmology AI applications.

Contribution

It proposes a novel disentanglement loss based on distance correlation to separate confounding factors in retinal images, enhancing interpretability and generation control.

Findings

Effective disentanglement of patient and camera attributes

Improved controllable and realistic image generation

Validated through qualitative and quantitative analyses

Abstract

Retinal fundus images play a crucial role in the early detection of eye diseases. However, the impact of technical factors on these images can pose challenges for reliable AI applications in ophthalmology. For example, large fundus cohorts are often confounded by factors like camera type, bearing the risk of learning shortcuts rather than the causal relationships behind the image generation process. Here, we introduce a population model for retinal fundus images that effectively disentangles patient attributes from camera effects, enabling controllable and highly realistic image generation. To achieve this, we propose a disentanglement loss based on distance correlation. Through qualitative and quantitative analyses, we show that our models encode desired information in disentangled subspaces and enable controllable image generation based on the learned subspaces, demonstrating the effectiveness of our disentanglement loss. The project's code is publicly available: https://github.com/berenslab/disentangling-retinal-images.