Fast Estimation of Specific Absorption Rate for Magnetic Resonance Imaging

Abstract

Specific absorption rate (SAR) is a main safety concern in magnetic resonance imaging (MRI). The radio frequency (RF) power deposition and SAR level have a significant impact on RF coil and pulse design. In high field MRI, multichannel RF transmission is widely used to mitigate B_1 field inhomogeneity. A specific B_1 map can be generated by adjusting the magnitude, phase and pulse shape of each coil. The dynamic changing of transmission will result in a subject dependent SAR distribution. Therefor we need to evaluate SAR in real subjects in real time. In addition, since the SAR is subject dependent, we need to investigate a large number of samples to get the statistical character of SAR. However, SAR evaluation requires full-wave electromagnetic simulation, which is time-consuming. So it is necessary to develop fast SAR evaluation methods. This dissertation presents some approaches to fast estimate the SAR in human body during MRI scan. Specifically, we present the following topics: (1) a brief introduction to finite-difference time-domain (FDTD) method is given and we verify the program by a scattering problem, a good agreement between analytical solution and simulation results is obtained; (2) a parallel implementation of FDTD based on GPU is provided and its performances and efficiency are compared with CPU; (3) SAR and temperature calculation are introduced. The B_1^+ field, SAR and temperature distribution within a human head in the birdcage coil at several frequencies as high as 498MHz are presented; (4) a hybrid technique that combines method of moments (MoM) and FDTD is presented to estimate local SAR rapidly, the accuracy and efficiency of this method are studied in detail; (5) a fast approach for statistical simulation of SAR based on the unscented transform is investigated, the number of sample points can be reduced and a meaningful safety margin can be established efficiently by this method. The methods presented in this dissertation can speed up the SAR estimation procedure, and they will be helpful to the coil design and SAR monitoring in MRI practice.