Generating Realistic Counterfactuals for Retinal Fundus and OCT Images using Diffusion Models

TL;DR

This paper introduces a diffusion model combined with robust classifiers to generate highly realistic counterfactual retinal images, aiding clinical decision explanations and disease understanding.

Contribution

It presents a novel approach for creating realistic retinal counterfactuals using diffusion models guided by disease-specific classifiers.

Findings

Counterfactuals are highly realistic and indistinguishable from real images.

Domain experts rated the generated images as more realistic than previous methods.

The method effectively depicts disease signs and removes lesions in retinal images.

Abstract

Counterfactual reasoning is often used in clinical settings to explain decisions or weigh alternatives. Therefore, for imaging based specialties such as ophthalmology, it would be beneficial to be able to create counterfactual images, illustrating answers to questions like "If the subject had had diabetic retinopathy, how would the fundus image have looked?". Here, we demonstrate that using a diffusion model in combination with an adversarially robust classifier trained on retinal disease classification tasks enables the generation of highly realistic counterfactuals of retinal fundus images and optical coherence tomography (OCT) B-scans. The key to the realism of counterfactuals is that these classifiers encode salient features indicative for each disease class and can steer the diffusion model to depict disease signs or remove disease-related lesions in a realistic way. In a user study, domain experts also found the counterfactuals generated using our method significantly more realistic than counterfactuals generated from a previous method, and even indistinguishable from real images.