计算机应用 ›› 2014, Vol. 34 ›› Issue (5): 1491-1493.DOI: 10.11772/j.issn.1001-9081.2014.05.1491

• 虚拟现实与数字媒体 • 上一篇    下一篇

压缩感知自动校准并行成像重建算法

张久明,郭树旭,王淼石,钟菲   

  1. 吉林大学 电子科学与工程学院，长春 130012
  • 收稿日期:2013-11-04 修回日期:2013-12-17 出版日期:2014-05-01 发布日期:2014-05-30
  • 通讯作者: 郭树旭
  • 作者简介:张久明（1990-），男，山东临沂人，硕士研究生，主要研究方向：医学图像处理；郭树旭（1959-），男，吉林长春人，教授，博士，主要研究方向：数字图像处理与分析、激光器噪声检测；王淼石（1988-），男，吉林长春人，博士研究生，主要研究方向：图像分割；钟菲（1983-），女，吉林长春人，博士研究生，主要研究方向：无线通信。

Reconstruction algorithm for auto-calibrating parallel imaging and compressed sensing

ZHANG Jiuming,GUO Shuxu,WANG Miaoshi,ZHONG Fei   

  1. College of Electronic Science and Engineering, Jilin University, Changchun Jilin 130012, China
  • Received:2013-11-04 Revised:2013-12-17 Online:2014-05-01 Published:2014-05-30
  • Contact: GUO Shuxu

PDF

摘要：

针对核磁共振并行成像重建提出了一种联合稀疏性模型，并与新的软阈值函数结合，将有助于提高重建图像质量。首先利用校准数据生成重建核，重建未采样数据点；然后采用联合稀疏性模型和新的软阈值函数，对各线圈图像数据进行处理；最后用改进的凸投影集算法（POCS）对压缩感知核磁共振并行成像进行重建。对于仿真图像和脑部图像，改进算法相比原算法，重建图像归一化均方根误差(nRMSE)在加速比为4时分别减少了23%和9%。实验结果表明，加速比较大时改进算法能明显提高并行成像重建图像的准确性。

Abstract:

A joint sparsity model, combined with a new soft threshold function, for magnetic resonance parallel imaging reconstruction was presented to improve the quality of reconstructed images. Firstly, the calibration data was used to generate the reconstruction kernel, and the non-acquired data points were reconstructed. Then the joint sparsity model and a new soft threshold function were used to process the image data for each coil. Finally, the compressed sensing magnetic resonance parallel imaging was reconstructed by the improved Projection Over Convex Set (POCS) algorithm. For the simulation images and brain images, the improved algorithm was compared with the original algorithm. When the acceleration ratio was 4, the normalized Root Mean Squared Error (nRMSE) of the reconstructed images was respectively reduced by 23% and 9%. The experimental results show that the improved algorithm can significantly improve the accuracy of the reconstructed parallel imaging images when the acceleration ratio is large relatively.

中图分类号: 

版权所有 © 2021 四川计算机应用杂志社有限公司
蜀ICP备 05010208 号-3　新出网证(川)字026号
技术支持：北京玛格泰克科技发展有限公司
访问总数：
今日访问：