Novel techniques regarding specific absorption rate and field of view reduction in magnetic resonance imaging

Abstract

In this dissertation, novel strategies regarding the reduction of the specific absorption rate (SAR) and the reduction of the total scan time, and analytic calculation methods for the lower limit on the specific absorption rate and the upper limit on the signal-to-noise ratio (SNR) are proposed. The first contribution of this dissertation is on the ultimate intrinsic signal-to-noise ratio (UISNR) and the ultimate intrinsic specific absorption rate (UISAR). Analytic expressions that are valid for arbitrarily shaped subjects are derived for these two parameters at the quasi-static limit. By comparing the UISNR expression to a previously published semi-analytic method for a cylindrical subject, it is shown that the maximum error is below 10%. In the primary contribution of this dissertation, gradient fields with nonlinear variation in space are used for radio-frequency (RF) excitation pulse design. When such fields are used for a pulse design, the relation between the excitation profile and the RF pulse is altered, which leads to a different RF envelope and hence, a different SAR value. Using simulations and experiments, SAR reductions between 15% - 54% are demonstrated, in three case studies. Another topic of this dissertation is on the reduction of the total scan time using nonlinear gradient fields. It is demonstrated that, when nonlinear gradient fields are used for excitation, the excitation region can be focused along more than a single direction. Furthermore, with a careful selection of readout encoding direction, reduced field-of-view imaging can be made without changing the SAR or the echo time. In a volunteer experiment, 60% reduction in the total scan time is obtained. The last topic of this dissertation is on curved slice imaging. It is shown using simulations that RF encoding can be used for imaging a curved field-of-view with non-rectangular and nonuniform voxels that may conform to the region-of-interest better. It is proposed that when the method is used with multi-dimensional excitation pulses, curved regions may be imaged in a shorter time