Convolutional Analysis Operator Learning by End-To-End Training of Iterative Neural Networks

TL;DR

This paper introduces a method for learning convolutional sparsifying filters through end-to-end training of iterative neural networks, improving image reconstruction in MRI by integrating the physical model into the learning process.

Contribution

It presents a novel approach to learn sparsifying filters directly within the reconstruction process, considering the physical model, unlike traditional decoupled pre-training methods.

Findings

Learned filters outperform pre-trained ones in MRI reconstruction

End-to-end training improves reconstruction quality

Applicable to non-Cartesian 2D cardiac cine MRI

Abstract

The concept of sparsity has been extensively applied for regularization in image reconstruction. Typically, sparsifying transforms are either pre-trained on ground-truth images or adaptively trained during the reconstruction. Thereby, learning algorithms are designed to minimize some target function which encodes the desired properties of the transform. However, this procedure ignores the subsequently employed reconstruction algorithm as well as the physical model which is responsible for the image formation process. Iterative neural networks - which contain the physical model - can overcome these issues. In this work, we demonstrate how convolutional sparsifying filters can be efficiently learned by end-to-end training of iterative neural networks. We evaluated our approach on a non-Cartesian 2D cardiac cine MRI example and show that the obtained filters are better suitable for the corresponding reconstruction algorithm than the ones obtained by decoupled pre-training.