Parallel transmission for ultrahigh-field imaging

Abstract

The development of MRI systems operating at or above 7 T has provided researchers with a new window into the human body, yielding improved imaging speed, resolution and signal-to-noise ratio. In order to fully realise the potential of ultrahigh-field MRI, a range of technical hurdles must be overcome. The non-uniformity of the transmit field is one of such issues, as it leads to non-uniform images with spatially varying contrast. Parallel transmission (i.e. the use of multiple independent transmission channels) provides previously unavailable degrees of freedom that allow full spatial and temporal control of the radiofrequency (RF) fields. This review discusses the many ways in which these degrees of freedom can be used, ranging from making more uniform transmit fields to the design of subject-tailored RF pulses for both uniform excitation and spatial selection, and also the control of the specific absorption rate. © 2015 The Authors. NMR in Biomedicine published by John Wiley & Sons Ltd.

Keywords: B1 mapping; RF shimming; SAR; parallel transmission; ultrahigh-field MRI.

Figure 1

Illustrative example of B ₁ ⁺ shimming. (A) Net transmit sensitivities produced by a 7T, eight‐channel dipole array when transmitting with default drives (top left), phase shimming (centre left) and maximum efficiency (bottom left). The prostate is indicated by the red box. The average sensitivities in the regions of interest (ROIs) are given by ${\overline{S}}_{\mathrm{R}\mathrm{O}\mathrm{I}}$. The weights obtained for each method are given on the right. (B) Net transmit sensitivities produced by a 7T, 12‐channel transverse electromagnetic (TEM) array with default drives (left), static parallel transmission (PTx) weights which maximise the minimum B ₁ ⁺ (centre), and applying weights which minimise the coefficient of variation (right). The minimum sensitivity in each slice is given by S _min. (Data courtesy of Dr Alessandro Sbrizzi, Dr Alexander Raaijmakers and Dr Hans Hoogduin, UMC Utrecht, the Netherlands).

Figure 2

k‐space trajectories for flip angle shimming. Black lines indicate the path through k‐space, and shaded regions indicate where radiofrequency (RF) transmission occurs, with the colour of the shading indicating the k‐space velocity at that point. The top row shows trajectories which are spatially selective in a single direction and, consequently, much higher spatial frequencies are visited in that dimension.

Figure 3

(A) L‐curve showing the trade‐off between power and homogeneity for a five‐spoke flip angle shimming pulse solved using linear least squares (LLS) and magnitude least squares (MLS). The best MLS B ₁ ⁺ shimming result is also shown. (B) Excitation magnitudes (top), excitation phases (centre) and the magnitude error (bottom) with respect to the target excitation of 0.5. The displayed spokes solutions were chosen to have the same power as the best MLS B ₁ ⁺ shimming solution. (Data courtesy of Dr Alessandro Sbrizzi and Dr Hans Hoogduin, UMC Utrecht, the Netherlands.)

Figure 4

Liver imaging at 7 T with increasing numbers of spokes. Gradient echo images were each obtained within a single breath hold. In each case, homogeneity was optimised over the liver only and the images were processed to remove the receive field profiles. Homogeneity in the liver improves with an increasing number of spokes. (Images courtesy of Dr Xiaoping Wu, University of Minnesota, MN, USA, originally from Quant. Imaging Med. Surg. 2014; 4: 4–10, with permission.)

Figure 5

Matched coronal slices from T ₁‐weighted magnetisation‐prepared rapid gradient echo (MP‐RAGE) images acquired at 7 T using standard non‐selective and spiral non‐selective (SPINS) radiofrequency (RF) pulses with a two‐channel head transmit coil. Both were acquired at an isotropic resolution of 0.8 mm with a flip angle of 8°, shot interval of 3.5 s, inversion delay of 1.2 s, TR = 9 ms, TE = 2.9 ms, parallel imaging reduction factors of 1.3x2 (anterior‐posterior x right‐left) with a total imaging time of approximately 10 minutes in both cases. Image uniformity was seen to have improved throughout the head, particularly in the cerebellum, as shown here. (Images courtesy of Dr Hans Hoogduin, UMC Utrecht, the Netherlands.)

Figure 6

Three‐dimensional eight‐channel parallel transmission‐local excitation (PTx‐LEx) pulse design for three‐dimensional (3D) fast spin echo imaging at 3 T 116. The gradient waveforms (top left) result in a 3D shells excitation k‐space trajectory [top right, ref. 221] consisting of multiple nested shells that are coloured separately here for clarity. The overall pulse duration is 12.3 ms. The pulses (middle left, different colours indicate different channels) are designed to produce a 90° excitation in the target volume placed over the cerebellum (middle right). Full field of view (FOV) image using non‐selective hard pulse excitation (bottom left; isotropic resolution of 1 mm with parallel imaging reduction factors of 1.7x1.7 (anterior‐posterior x right‐left), full FOV image using designed LEx (bottom middle) and reduced FOV image using LEx (bottom right, isotropic resolution of 0.8 mm without parallel imaging).

Figure 7

T ₂‐weighted three‐dimensional fast spin echo (FSE) imaging at 7 T using dynamically modulated k _T‐points radiofrequency (RF) pulses for excitation and refocusing. The diagram (bottom) depicts the k _T‐points RF pulses used, consisting of multiple hard pulses. The amplitudes and phases of these hard pulses are optimised so that, during each shot of the FSE sequence, the magnetization is brought to a pseudo‐steady state (PSS) with desired echo amplitude by the first P₁ pulse (here P₁ = 10), and then subsequently maintained in this state, despite the presence of strong B ₁ ⁺ non‐uniformity. The spatially resolved extended phase graph (SR‐EPG) framework is used to predict the echo amplitudes for all locations in space and at each TE, and these are optimized to be uniform 182. Dynamic modulation allows more uniform signals to be obtained, recovering reduced signals that are apparent in the temporal lobes (see increased signal apparent on the ratio image). (Images courtesy of Dr Florent Eggenschwiler, CIBM, Lausanne, Switzerland.)

Figure 8

Specific absorption rate (SAR) reduction achieved by combining a loop and dipole array. The dipoles are orthogonal to the main magnetic field (z‐axis), and therefore produce radiofrequency (RF) magnetic fields primarily in the z‐axis – these produce very little contribution to the NMR and are hence referred to as ‘dark modes’. The electric fields produced by the dipoles can, however, be used to cancel those produced by the loops, thus reducing local SAR. In this example, magnitude least squares (MLS) B ₁ ⁺ shimming using an array of four loops and four dipoles was able to reduce peak 10 gram SAR by 36% when compared with a four‐loop linear phased array, whilst producing an almost identical B ₁ ⁺. (Images courtesy of Dr Yigitcan Eryaman, University of Minnesota, MN, USA. Data reproduced from Magn. Reson. Med, Epub 2014, doi: 10.1002/mrm.25246., with permission).