Skip to main navigation Skip to search Skip to main content

Cylindrical-surface shim coil design using target-field method combined with fast L1 norm minimization algorithm

  • Geli Hu
  • , Zhipeng Ni
  • , Qiuliang Wang
  • CAS - Institute of Electrical Engineering

Research output: Contribution to journal › Article › peer-review

1 Scopus citations

Abstract

The target field (TF) method proposed by Turner in 1986 has become the mainstream method for both shim coil and gradient coil in magnetic resonance imaging (MRI). A series of methods based on the original one have been proposed. In order to solve the problem of finite size during coil design, Forbes and Crozier proposed a novel approach which employs the trigonometric functions to expand surface current density. To overcome the ill-posed problem in the equation solving process, the undetermined coefficients are estimated by the combination of the standard least square (LS) with L2 norm regularization (LS-L2), and good design results are got. This paper proposed to estimate the coefficients of the current density using LS with L1 norm regularization (LS-L1). This new method was employed to design a set of shim coils for an MRI system. The length of the main magnet used in this system was 3 m; the diameter of room temperature and region of interest (ROI) was 1.6 m, 0.4 m, respectively. Simulation results via Matlab demonstrate that the proposed LS-L1 method performed better than the traditional LS-L2 TF method.

Original languageEnglish
Pages (from-to)4737-4742
Number of pages6
JournalZhongguo Dianji Gongcheng Xuebao/Proceedings of the Chinese Society of Electrical Engineering
Volume34
Issue number27
DOIs
StatePublished - 25 Sep 2014

Keywords

  • L norm
  • L norm
  • Least square (LS)
  • Magnetic resonance imaging
  • Shim coils
  • Target field method

Fingerprint

Dive into the research topics of 'Cylindrical-surface shim coil design using target-field method combined with fast L1 norm minimization algorithm'. Together they form a unique fingerprint.

Cite this