Synopsis
High field MRI suffers from non-uniform transmit fields and B0
variations due to increased susceptibility effects, making uniform slice-excitation
very difficult. We developed a new pulse trajectory – the “twisted spokes” RF pulse
– to achieve accurate slice-selection with high in-plane uniformity and greatly
improved B0 robustness. The twisted spokes trajectory consists of helical k-space segments
oriented along the slice-selection direction (e.g., kz). We found
that, when the helical segments are designed appropriately, the resulting RF
pulses are short, achieve sharp slice profiles and uniform flip-angle
distributions, and – at the same time – are very robust to off-resonance
effects.Target audience
RF engineers, MR physicists, ultra-high field practitioners.
Purpose
The potential benefits of ultra-high field MRI – higher resolution and
sensitivity – are, in practice, hard to realize because of non-uniformities
of the transmit fields (B
1+) and susceptibility effects that distort the image and lead to through-slice signal dropout. A popular
strategy to create uniform flip-angle (FA) distributions using a
slice-selective sequence is the spokes pulse [1,2,3]. Using parallel
transmission (pTx), the complex weightings of the spokes played on each
transmit channel can be optimized to maximize FA uniformity within the slice.
Spokes pulses achieve this by sampling k
z (slice direction) at high
resolution and by depositing energy at discrete locations in the k
x-k
y
plane. However, spokes pulses are extremely sensitive to off-resonance effects
so that, in practice, their performance worsens dramatically close to the sinus
and ear canal cavities [4]. This is due to the fact that spokes pulses are
relatively long and deposit RF energy at discrete locations, so that errors in
the position of the spokes in the k
x-k
y plane (e.g., due
to B
0 effects) lead to inaccurate excitations. In this work, we introduce a
new pulse trajectory that we call the “twisted spokes”. This pulse is able to
excite sharp slice-selection profiles while creating highly uniform FA distributions
at 7T. Importantly, the pulse is very robust to off-resonance effects, because the pulse duration is kept short and the RF energy is not deposited at
discrete k
x-k
y locations but is instead distributed
across the k
x-k
y plane.
Methods
Gradient and RF Waveforms: The twisted spokes trajectory consists
of two helical k-space segments oriented along the k
z-axis (
z being the slice direction, Figure 1).
These segments have a large extent along k
z (k
z,max=150 m
-1,
Δz=6.0 mm) to ensure sharp
slice-selection and only a minor extent in k
x/y (≈10 m
-1).
The exact shape of the twisted spokes segments (e.g., extent along the k-space axes,
number of revolutions, etc.) has a large impact on the excitation quality. We
control these basic features of the twisted spokes by so-called shape
parameters that we optimize using our joint k-space trajectory and RF waveform
design method [5]. In this method, we optimize the shape parameters for a
given target profile (
z-slice) and given B
0/B
1+ maps using a nested optimization approach in which
the inner loop is a small tip-angle RF pulse design on a constant k-space
trajectory. The outer loop is a constrained optimization on the shape
parameters. In order to achieve sharp and uniform slice excitation and B
0
robustness at the same time, we optimize the RF waveforms (inner loop) such
that the target FA map is realized simultaneously at offset frequencies −50 Hz,
0 Hz, +50 Hz [6,7]. To improve the excitation quality, the last RF design is a
magnitude least-squares optimization that does not impose a predetermined spin
phase.
Evaluation: Our RF pulses were evaluated using Bloch simulations
based on B
0 and B
1+ maps (8 channels) acquired
on a 7T scanner (“Step 2” pTx, Magnetom 7T, Siemens, Erlangen) loaded with a
realistic 3D-printed head phantom with three compartments (bone, brain and
everything else) [4]. The target was to excite a 6 mm z-slice at 10° FA in about 1.5 ms with good performance regarding in-plane uniformity in the brain (uniformity in the skull is not considered), through-plane slice profile, and robustness to off-resonance effects. Our twisted
spokes pulses are compared to four different spokes strategies: (
A) 1-spoke (birdcage mode), (
B)
1-spoke (MLS optimization), (
C) 2-spokes
(MLS, no B
0 robustness), and (
D) 2-spokes
(MLS, B
0 robustness enforced as explained in [4,7]). All pulses satisfy
the gradient system (G
max = 40 mT/m, S
max = 150 mT/m) and
peak RF power (V
max = 150 V per channel) constraints.
Results
In Figure 2, the results of the six RF pulses are shown. The 1-spoke birdcage
mode pulse (
A) exhibits severe in-plane
FA variations which is partly compensated by optimizing the complex weights of
each channel (
B). Both 1-spoke
pulses are intrinsically robust to B
0 effects. The 2-spokes pulses
are rather long (~2.6 ms) and either achieve in-plane FA uniformity (
C) or off-resonance robustness (
D), but not both. Our optimized twisted
spokes pulse (
F) is short (1.8 ms)
and produces a uniform in-plane FA and a good slice profile. At the same time
and compared to both 2-spokes pulses, B
0 robustness is dramatically
improved. Comparing the non-optimized (
E)
and optimized (
F) twisted spokes RF
pulses emphasizes the impact of trajectory shape optimization on FA fidelity, B
0 robustness, as well as average power (P
mean reduced from 27.5 W to
5.3 W).
Acknowledgements
No acknowledgement found.References
[1] Saekho et al., MRM 53(2), 2005; [2] Saekho et al., MRM 55(4), 2006;
[3] Setsompop et al., MRM 60(6), 2008; [4] Guérin et al., MRM 2015, in press;
[5] Davids et al., MRM 2015, in press; [6] Grissom et al., MRM 56(3), 2006; [7]
Setsompop et al., MRM 61(2), 2009