Share Email Print

Proceedings Paper

Parallel and accelerated computing method based on echo-memory GPGPU for virtual MRI scanning
Author(s): Bin Kuang; Nini Rao; Dingyun Liu; Yongli Wan; Chaoming He
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

MRI simulator is a very useful tool to study and learn the MRI. However, most of MRI simulators are very time-consuming due to their grid computing frameworks. Although some parallel computing approaches have been used to improve their computational complexity, the efficiency is still far from satisfactory. In this paper, a novel GPGPU based parallel computing accelerated Bloch solver is presented to solve this problem. In the proposed method, the models including virtual object, inhomogeneity of static field, virtual sequence controller and parallelized magnetization computation kernel are optimized and designed to meet the requirement of the parallel computing framework given by the GPGPU. In addition, to balance the memory complexity which is introduced by the GPGPU and the parallel computing performance, the echo memory mode is designed to reduce the huge memory required by the GPU device. This solution can fit the MR scanning in a more natural way than the traditional pipe-line streams design. Several tests confirmed that high scale simulation can be run smoothly with high acceleration factor. The proposed method can effectively accelerate the MRI simulator under the acceptable performances.

Paper Details

Date Published:
Proc. SPIE 9412, Medical Imaging 2015: Physics of Medical Imaging, 94122M; doi: 10.1117/12.2081780
Show Author Affiliations
Bin Kuang, Univ. of Electronic Science and Technology of China (China)
Siemens Magnetic Resonance Ltd. (China)
Nini Rao, Univ. of Electronic Science and Technology of China (China)
Dingyun Liu, Univ. of Electronic Science and Technology of China (China)
Yongli Wan, Univ. of Electronic Science and Technology of China (China)
Chaoming He, Siemens Magnetic Resonance Ltd. (China)

Published in SPIE Proceedings Vol. 9412:
Medical Imaging 2015: Physics of Medical Imaging
Christoph Hoeschen; Despina Kontos, Editor(s)

© SPIE. Terms of Use
Back to Top