Deformation-aware GAN for Medical Image Synthesis with Substantially Misaligned Pairs

TL;DR

This paper introduces DA-GAN, a novel deformation-aware GAN that dynamically corrects substantial misalignments during medical image synthesis, improving accuracy and image quality in challenging scenarios like respiratory motion artifacts.

Contribution

The paper proposes a deformation-aware GAN with multi-objective inverse consistency and deformation-aware discriminators to handle large misalignments in medical image synthesis, advancing beyond existing registration-based methods.

Findings

DA-GAN outperforms existing methods on simulated misalignments

Achieves superior image fidelity on lung MRI-CT data with motion artifacts

Demonstrates potential for clinical applications like radiotherapy planning

Abstract

Medical image synthesis generates additional imaging modalities that are costly, invasive or harmful to acquire, which helps to facilitate the clinical workflow. When training pairs are substantially misaligned (e.g., lung MRI-CT pairs with respiratory motion), accurate image synthesis remains a critical challenge. Recent works explored the directional registration module to adjust misalignment in generative adversarial networks (GANs); however, substantial misalignment will lead to 1) suboptimal data mapping caused by correspondence ambiguity, and 2) degraded image fidelity caused by morphology influence on discriminators. To address the challenges, we propose a novel Deformation-aware GAN (DA-GAN) to dynamically correct the misalignment during the image synthesis based on multi-objective inverse consistency. Specifically, in the generative process, three levels of inverse consistency cohesively optimise symmetric registration and image generation for improved correspondence. In the adversarial process, to further improve image fidelity under misalignment, we design deformation-aware discriminators to disentangle the mismatched spatial morphology from the judgement of image fidelity. Experimental results show that DA-GAN achieved superior performance on a public dataset with simulated misalignments and a real-world lung MRI-CT dataset with respiratory motion misalignment. The results indicate the potential for a wide range of medical image synthesis tasks such as radiotherapy planning.