A numerical study of the field dependence of signal-to-noise ratio in high-field MRI

The increasing interest in the use of magnetic resonance imaging (MRI) for quantitative spectroscopy, real-time imaging, magnetic resonance angiography (MRA), and other emerging applications has led to increased effort towards maximizing the available signal-to-noise ratio (SNR). For a designer to be able to maximize the available SNR, an accurate analysis method is required. However, many of the trends in hardware design intended to increase SNR (such as the move towards higher B/sub 0/ field strengths) make the analysis of SNR more complicated. The intrinsic SNR (ISNR) is a quantity which represents a fundamental limit of the performance of an MR system for a given patient/coil configuration. We use the finite-difference time-domain (FDTD) method to study the SNR of MRI at frequencies up to 340 MHz (corresponding to the Larmour frequency of protons in an 8 T B/sub 0/ field). The principle of reciprocity together with the FDTD analysis is used to provide a means of computing the intensity of the NMR signal from all points in the sample. The FDTD can accommodate an inhomogeneous model of the patient being imaged, which more accurately describes the physical situation than previous approaches. This technique is used to assess the field-dependence of the ISNR in high-field MRI.