Deformation-Compensated Learning for Image Reconstruction without Ground Truth

TL;DR

This paper introduces DeCoLearn, a novel method for training deep image reconstruction networks without ground-truth images, by compensating for nonrigid deformations using a jointly trained registration module, validated on MRI data.

Contribution

DeCoLearn is the first approach to enable learning from measurements of deforming objects without ground-truth, combining deep registration with reconstruction in a unified framework.

Findings

Significantly improves MRI image quality in simulated and real data.

Effectively compensates for nonrigid deformations during training.

Outperforms existing methods that lack deformation compensation.

Abstract

Deep neural networks for medical image reconstruction are traditionally trained using high-quality ground-truth images as training targets. Recent work on Noise2Noise (N2N) has shown the potential of using multiple noisy measurements of the same object as an alternative to having a ground-truth. However, existing N2N-based methods are not suitable for learning from the measurements of an object undergoing nonrigid deformation. This paper addresses this issue by proposing the deformation-compensated learning (DeCoLearn) method for training deep reconstruction networks by compensating for object deformations. A key component of DeCoLearn is a deep registration module, which is jointly trained with the deep reconstruction network without any ground-truth supervision. We validate DeCoLearn on both simulated and experimentally collected magnetic resonance imaging (MRI) data and show that it significantly improves imaging quality.