A comparison of MRI magnet design using a Hopfield network and the optimized material distribution method

Author

Lowther, D.A. ; Mai, W. ; Dyck, D.N.

Author_Institution

Dept. of Electr. Eng., McGill Univ., Montreal, Que., Canada

Volume

34

Issue

5

fYear

1998

fDate

9/1/1998 12:00:00 AM

Firstpage

2885

Lastpage

2888

Abstract

Two approaches to device design, the Hopfield network and optimized material distribution (OMD), are compared. The comparison is based on a previously published MRI dipole magnet design problem. Since this problem is already formulated as a material distribution problem, the results focus on the optimization procedures used in the two methods, i.e. the learning algorithm of the Hopfield network versus the conjugate gradient algorithm used in OMD. Results of the two methods are presented, and are also compared to a previously published simulated annealing solution

Keywords

Hopfield neural nets; conjugate gradient methods; electromagnets; learning (artificial intelligence); magnetic resonance spectroscopy; optimisation; Hopfield network; MRI magnet; conjugate gradient algorithm; learning algorithm; magnet design; optimized material distribution method; Boundary conditions; Coils; Design optimization; Hopfield neural networks; Magnetic materials; Magnetic resonance imaging; Neurons; Optimization methods; Simulated annealing; Testing;

fLanguage

English

Journal_Title

Magnetics, IEEE Transactions on

Publisher

ieee

ISSN

0018-9464

Type

jour

DOI

10.1109/20.717672

Filename

717672