Research of MRI Reconstruction Method by Using De-aliasing Wasserstein Generative Adversarial Networks with Gradient Penalty

YUAN Zi-han; JIANG Ming-feng; LI Yang; ZHI Ming-hao; ZHU Zhi-jun

doi:10.3969/j.issn.0372-2112.2020.10.002

您当前的位置：

首页 >

文章列表页 >

Research of MRI Reconstruction Method by Using De-aliasing Wasserstein Generative Adversarial Networks with Gradient Penalty

更新时间：2023-02-24

- Research of MRI Reconstruction Method by Using De-aliasing Wasserstein Generative Adversarial Networks with Gradient Penalty
- Acta Electronica Sinica Vol. 48, Issue 10, Pages: 1883-1890(2020)
- 作者机构：
  
  1. 浙江理工大学信息学院,浙江,杭州,310018
  3. 解放军第一一七医院心血管内科,浙江,杭州,310013
- 作者简介：
- 基金信息：
- DOI：10.3969/j.issn.0372-2112.2020.10.002
  CLC： TP302
- Published Online：25 October 2020，
  
  Published：2020
- 稿件说明：
移动端阅览
YUAN Zi-han, JIANG Ming-feng, LI Yang, et al. Research of MRI Reconstruction Method by Using De-aliasing Wasserstein Generative Adversarial Networks with Gradient Penalty[J]. Acta Electronica Sinica, 2020, 48(10): 1883-1890.
DOI：

YUAN Zi-han, JIANG Ming-feng, LI Yang, et al. Research of MRI Reconstruction Method by Using De-aliasing Wasserstein Generative Adversarial Networks with Gradient Penalty[J]. Acta Electronica Sinica, 2020, 48(10): 1883-1890. DOI： 10.3969/j.issn.0372-2112.2020.10.002.

摘要

本文提出了一种基于改进Wasserstein生成式对抗网络（De-aliasing Wasserstein Generative Adversarial Network with Gradient Penalty，DAWGAN-GP）的磁共振图像重构算法，该方法利用Wasserstein生成式对抗网络代替传统的生成式对抗网络，并结合梯度惩罚的方法提高训练速度，解决WGAN收敛缓慢问题.此外，为了有更好的重构效果，我们将感知损失，像素损失和频域损失引入至损失函数中进行网络训练.实验结果表明，对比现有的基于深度学习的磁共振图像重构算法，基于DAWGAN-GP的磁共振图像重构方法具有更好的重构效果，可获得更高的峰值信噪比（Peak Signal to Noise Ratio，PSNR）和更好的结构相似性（Structural Similarity Index Measure，SSIM）.

Abstract

In this paper

we propose an improved Wasserstein generative adversarial network (WGAN)

de-aliasing Wasserstein generative adversarial network with Gradient Penalty (DAWGAN-GP)

for magnetic resonance imaging (MRI) reconstruction.This method uses WGAN to replace the traditional GAN

and combined the gradient penalty to improve the training speed and to solve the slow convergence problem of WGAN.In addition

for better preservation of the fine structures in the reconstructed images

we incorporate perceptual loss

pixel loss and frequency loss into the loss function for training the network.Compared with other state-of-the-art deep learning methods for MR images reconstruction

DAWGAN-GP method outperforms all other methods and can provide superior reconstruction with improved peak signal to noise ratio (PSNR) and better structural similarity index measure (SSIM).

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Feature-Space Optimization-Inspired and Self-Attention Enhanced Neural Network Reconstruction Algorithm for Image Compressive Sensing

Sparse Reconstruction for Omnidirectional Image Based on Total Variation

Joint Image Registration and Pattern-Based High-Resolution Image Reconstruction

Image Compressive Sensing Reconstruction Network Based on Iterative SPL Theory

A Three Order Hybrid Algorithm for Electrical Capacitance Tomography

Related Author

CHEN Wen-jun

YANG Chun-ling

ZHANG Mao-jun

WANG Wei

TAN Shu-ren

LI Yong-le

LOU Jing-tao

PENG Hong

Related Institution

School of Electronic and Information Engineering, South China University of Technology

College of Information System and Management, National University of Defense Technology

College of Computer Science and Engineering,South China University of Technology

Department of Computer Science,Shaoguan University

College of Computer Science and EngineeringSouth China University of TechnologyGuangzhouGuangdong 510641China

⁰