Steerable Conditional Diffusion for Out-of-Distribution Adaptation in Medical Image Reconstruction

TL;DR

This paper introduces Steerable Conditional Diffusion, a new sampling framework that adapts diffusion models during reconstruction to improve out-of-distribution performance in medical imaging, reducing hallucinations and enhancing accuracy.

Contribution

The paper proposes a novel adaptive sampling method that adjusts diffusion models during reconstruction based on measurements, addressing out-of-distribution challenges in medical imaging.

Findings

Significant improvement in out-of-distribution reconstruction quality

Reduction of hallucinated features in generated images

Enhanced robustness across diverse imaging modalities

Abstract

Denoising diffusion models have emerged as the go-to generative framework for solving inverse problems in imaging. A critical concern regarding these models is their performance on out-of-distribution tasks, which remains an under-explored challenge. Using a diffusion model on an out-of-distribution dataset, realistic reconstructions can be generated, but with hallucinating image features that are uniquely present in the training dataset. To address this discrepancy during train-test time and improve reconstruction accuracy, we introduce a novel sampling framework called Steerable Conditional Diffusion. Specifically, this framework adapts the diffusion model, concurrently with image reconstruction, based solely on the information provided by the available measurement. Utilising our proposed method, we achieve substantial enhancements in out-of-distribution performance across diverse imaging modalities, advancing the robust deployment of denoising diffusion models in real-world applications.