Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SRE-CNN: A Spatiotemporal Rotation-Equivariant CNN for Cardiac Cine MR Imaging

Yuliang Zhu, Jing Cheng, Zhuo-Xu Cui, Jianfeng Ren, Chengbo Wang, Dong Liang

TL;DR

This work proposes a novel framework of Spatiotemporal Rotation-Equivariant CNN (SRE-CNN), spanning from the underlying high-precision filter design to the construction of the temporal-equivariant convolutional module and imaging model, to fully harness the rotation symmetries inherent in dynamic MR images.

Abstract

Dynamic MR images possess various transformation symmetries,including the rotation symmetry of local features within the image and along the temporal dimension. Utilizing these symmetries as prior knowledge can facilitate dynamic MR imaging with high spatiotemporal resolution. Equivariant CNN is an effective tool to leverage the symmetry priors. However, current equivariant CNN methods fail to fully exploit these symmetry priors in dynamic MR imaging. In this work, we propose a novel framework of Spatiotemporal Rotation-Equivariant CNN (SRE-CNN), spanning from the underlying high-precision filter design to the construction of the temporal-equivariant convolutional module and imaging model, to fully harness the rotation symmetries inherent in dynamic MR images. The temporal-equivariant convolutional module enables exploitation the rotation symmetries in both spatial and temporal dimensions, while the high-precision convolutional filter, based on parametrization strategy, enhances the utilization of rotation symmetry of local features to improve the reconstruction of detailed anatomical structures. Experiments conducted on highly undersampled dynamic cardiac cine data (up to 20X) have demonstrated the superior performance of our proposed approach, both quantitatively and qualitatively.

SRE-CNN: A Spatiotemporal Rotation-Equivariant CNN for Cardiac Cine MR Imaging

TL;DR

This work proposes a novel framework of Spatiotemporal Rotation-Equivariant CNN (SRE-CNN), spanning from the underlying high-precision filter design to the construction of the temporal-equivariant convolutional module and imaging model, to fully harness the rotation symmetries inherent in dynamic MR images.

Abstract

Dynamic MR images possess various transformation symmetries,including the rotation symmetry of local features within the image and along the temporal dimension. Utilizing these symmetries as prior knowledge can facilitate dynamic MR imaging with high spatiotemporal resolution. Equivariant CNN is an effective tool to leverage the symmetry priors. However, current equivariant CNN methods fail to fully exploit these symmetry priors in dynamic MR imaging. In this work, we propose a novel framework of Spatiotemporal Rotation-Equivariant CNN (SRE-CNN), spanning from the underlying high-precision filter design to the construction of the temporal-equivariant convolutional module and imaging model, to fully harness the rotation symmetries inherent in dynamic MR images. The temporal-equivariant convolutional module enables exploitation the rotation symmetries in both spatial and temporal dimensions, while the high-precision convolutional filter, based on parametrization strategy, enhances the utilization of rotation symmetry of local features to improve the reconstruction of detailed anatomical structures. Experiments conducted on highly undersampled dynamic cardiac cine data (up to 20X) have demonstrated the superior performance of our proposed approach, both quantitatively and qualitatively.
Paper Structure (10 sections, 5 equations, 3 figures, 1 table)

This paper contains 10 sections, 5 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Method
  3. Accelerated MRI Reconstruction Model
  4. Spatiotemporal Rotation-Equivariant CNN Design
  5. Filter Parametrization
  6. Data and Experiments
  7. Results and Discussion
  8. Conclusion
  9. Acknowledgments.
  10. Disclosure of Interests.

Figures (3)

  • Figure 1: An example of 2D+t dynamic cardiac cine MRI data. Similar anatomical structures under different orientations are zoomed in and marked by colored bounding box.
  • Figure 2: (a) The unrolled reconstruction model with the PGD method. The layers in SRE-CNN are signed by different colors. (b) Illustration of an example network constructed by the proposed SRE-CNN, where we set the rotation transformation group $S ( A_i,B_i,\in S)$ as ${2\pi i}/3$ rotations, $i=1,2,3$. The proposed temporal equivariant layer, input and intermediate equivariant layer are marked by dashed lines. $p$ denotes the size of filter. $F$ denotes feature maps layer.
  • Figure 3: Reconstructions in the spatial domain and along the temporal dimension (x-t image) under acceleration (R=12 and 20) and the corresponding error maps.