Transform learning MRI with global wavelet regularization

Abstract

Sparse regularization of the reconstructed image in a transform domain has led to state of the art algorithms for magnetic resonance imaging (MRI) reconstruction. Recently, new methods have been proposed which perform sparse regularization on patches extracted from the image. These patch level regularization methods utilize synthesis dictionaries or analysis transforms learned from the patch sets. In this work we jointly enforce a global wavelet domain sparsity constraint together with a patch level, learned analysis sparsity prior. Simulations indicate that this joint regularization culminates in MRI reconstruction performance exceeding the performance of methods which apply either of these terms alone.

A Structurally Regularized Convolutional Neural Network for Image Classification Using Wavelet-Based Subband Decomposition

“…Several works focus on this approach. A jointly enforced global wavelet domain sparsity constraint together with a learned analysis sparsity prior was introduced in [14]. A wavelet-regularized semi-supervised learning algorithm using suitably defined spline-like graph wavelets was introduced in [15].…”

MRI reconstruction with joint global regularization and transform learning

“…In this paper we aim to combine the global, image-wide regularization as proposed in papers such as [3] or [7], with the effective patch-wise transform learning model of [27]. A preliminary version of our approach has been accepted for presentation in the EUSIPCO 2015 conference [28]. The presentation here is substantially revised and extended when compared to the conference version.…”