Page 260 - Biomedical Engineering and Design Handbook Volume 2, Applications

DESIGN OF MAGNETIC RESONANCE SYSTEMS 239

At higher field strengths, most of the RF losses are due to patients. However, for low field
systems, patient losses are much smaller than coil losses. One approach for reducing the SNR
impact of such low field coil losses is to use superconducting surface coils. 94 These coils are
typically limited in size and require attached cryostats. In addition, very high Q translates to
very narrow bandwidths and tight tolerances on tuning.

Receiver Coil Arrays. Signal-to-noise ratio may be improved by replacing a single, large field of
view coil with an array of smaller coils that combine for a field of view similar to the larger coil.
Each smaller coil receives less noise than the larger coil. Since noise is not correlated, while signals
are (due to coil design) signal to noise is typically higher when receiver coil arrays are used.
Typically, noise from adjacent coil elements is designed to be uncorrelated by appropriate overlap or
circuit design.

Parallel Imaging. Magnetic resonance parallel imaging techniques use multiple receiver coils
(with different spatial sensitivities). These receiver coils are used to supply some portion of the spatial
encoding information to reduce the total number of gradient phase encodings. 95–101 The techniques
typically are used to obtain shorter scan times and may shorten readout times. Some image artifacts
can be reduced using parallel imaging.
Typically, in parallel imaging receiver coil spatial sensitivity is determined from low-resolution
reference images. Then aliased data sets can be generated for each individual coil. Receiver coil
spatial sensitivity information may then be used with either image-domain or frequency (k-space)-
domain reconstruction techniques to generate full field of view (nonaliased) images.
Signal to noise in parallel imaging is reduced by the shorter imaging time and by nonideal coil
geometry. Nonideal coil geometry finds expression in the g factor. The g factor refers to the pixel-
by-pixel signal-to-noise ratio obtained with parallel imaging divided by the signal-to-noise ratio
using equivalent conventional techniques with the same receive coils. The g factor is only defined in
regions where signal to noise is above a threshold and technically applies only to image-based par-
allel imaging methods.
The reduction of the number of phase-codings needed due to multiple receiver coils in parallel
imaging is called the acceleration factor. Maximum acceleration factors are somewhat less than the
number of parallel imaging receiver coils.

8.4.4 Preamplifiers
Low-noise figure preamplifiers with noise impedances matched and relatively near the receiver
coils are required to avoid degrading SNR. Quadrature systems with preamplifiers on each channel
have small SNR advantages over quadrature systems employing low loss combiners and a single
preamplifier.

8.4.5 SAR

During MR scans below 3 T or so, RF power deposition in patients can be approximated from
quasistatic analysis, assuming electric field coupling to patients can be neglected. 10,34 Let R be the
radius, σ the conductivity, and ρ the density of a homogeneous, sphere of tissue (Fig. 8.1). Assume
that this sphere is placed in a uniform RF magnetic field of strength B and frequency ω. Let
1
the radiofrequency duty cycle be η. Then average specific absorption rate (SAR), SAR , may be
ave
expressed as 63
σηω 2 2 1 BR 2
SAR ave = (8.20)
20ρ