Dynamic MRI using Learned Transform-based Tensor Low-Rank Network (LT$^2$LR-Net)

TL;DR

This paper introduces LT$^2$LR-Net, a deep unrolling network for dynamic MRI that leverages learned transform-based tensor low-rank modeling to improve image reconstruction accuracy.

Contribution

It generalizes tensor SVD with an arbitrary unitary transform and integrates CNNs to adaptively learn transformations, enhancing tensor low-rank modeling in dynamic MRI reconstruction.

Findings

Outperforms state-of-the-art methods on cardiac cine MR data

Provides more accurate and robust image reconstructions

Demonstrates the effectiveness of learned transforms in tensor low-rank modeling

Abstract

While low-rank matrix prior has been exploited in dynamic MR image reconstruction and has obtained satisfying performance, tensor low-rank models have recently emerged as powerful alternative representations for three-dimensional dynamic MR datasets. In this paper, we introduce a novel deep unrolling network for dynamic MRI, namely the learned transform-based tensor low-rank network (LT$^2$LR-Net). First, we generalize the tensor singular value decomposition (t-SVD) into an arbitrary unitary transform-based version and subsequently propose the novel transformed tensor nuclear norm (TTNN). Then, we design a novel TTNN-based iterative optimization algorithm based on the alternating direction method of multipliers (ADMM) to exploit the tensor low-rank prior in the transformed domain. The corresponding iterative steps are unrolled into the proposed LT$^2$LR-Net, where the convolutional neural network (CNN) is incorporated to adaptively learn the transformation from the dynamic MR dataset for more robust and accurate tensor low-rank representations. Experimental results on the cardiac cine MR dataset demonstrate that the proposed framework can provide improved recovery results compared with the state-of-the-art methods.