Travelling-Wave Excitation for 16.4T Small

Travelling-Wave Excitation for 16.4T Small-Bore MRI
2
i
3
3
3
Patrick Bluem , Alexey Tonyushkin , Dinesh Deelchand , Gregor Adriany , Pierre-Francois Van de Moortele , Andrew J M
i
2
Kiruluta , Zoya Popovic
i
2
Abstract
-
University of Colorado, Boulder, Colorado
Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts
3
University of Minnesota Medical School, Minneapolis, Minnesota
In this paper, we present the design of a probe
for a travelling-wave 16.4T small-bore animal research MRI
system. The probe is a 698-MHz coaxially-fed microstrip patch
designed to give a circularly polarized magnetic field when
To date, a few pre-clinical human scanners at 7T and 9.4T
[3,9] and small bore research (animal) 16T and 21.lT
[10]
systems were used to demonstrate traveling-wave (TW) MRI.
placed in the bore cavity. Images of a water phantom using the
Whole-body scanner dimensions result in the bore being an
patch
simulations.
above cut-off cylindrical waveguide with one or two TW
Additionally, a periodic axial strip cylinder is inserted into the
modes propagating in the unloaded bore. In contrast, most of
probe
are
obtained
and
compared
with
bore, resulting in a 7-fold increase in SNR, and enabling both
gradient recalled echo and spin echo imaging of the phantom.
The modified mode content in the image is compared to full-wave
simulations.
Index Terms - Magnetic Resonance Imaging, waveguide,
travelling-wave, patch, circular polarization.
the small-bore research, as well as clinical «4T) systems, do
not support traveling waves unless high dielectric inserts [11]
or coaxial transmission lines [12] for mode propagation.
Typical RF probes for preclinical TW-MRI are electric dipole
probes or standard patch antennas that couple energy to the
loaded waveguide modes, with the goal of providing as
uniform as possible circularly polarized B-field inside the
I. INTRODUCTION
dielectric to be imaged [13].
In clinical 1.5T and 3T magnetic resonance imaging (MRI)
In this paper, we present the design of a travelling-wave
instruments, the object being imaged is closely coupled to the
probe for a 16.4 T small-bore MRI system, some bore
detector through near
modifications
fields
and detection is
performed
which
enable
increased
SNR,
and
finally
through Faraday probes [1,2]. Nuclear magnetic resonance
measured data on a water phantom obtained in the setup
(NMR) can also be excited and detected using long-range
shown in Figure 1.
coupling with travelling waves, demonstrated by several
research groups over the past few years, e.g. [3]. One benefit
76 em
of this approach is more uniform coverage of samples that are
100 em
Gradient coil region
larger than the wavelength of the NMR signal. Uniform spatial
coverage in MRI is traditionally achieved by tailoring the
26cm
reactive near field of resonant Faraday probes. This approach
Patch probe
is valid when the radio-frequency wavelength at the Larmor
frequency is substantially larger than the target volume, which
Phantom
12 em
!----- ---.---- 332 em --------- ---+
Magnet
does not hold for modern, wide-bore, high-field systems.
This paper addresses a new method of integrated design of
exposure
and
excitation
of
UHF
and
low
microwave
Fig. I.
Sketch of 16. T small bore Varrian MRl with relevant
dimensions and position of phantom shown. The traveling wave
frequency magnetic fields for next-generation MRI at high
probe position can be varied inside the wider cylinder to demonstrate
magnetic flux densities
exposure and detection without near-field coupling.
(Bo>
3 T) [4-6], which improve the
signal-to-noise ratio (SNR) [7]. For example, at 7 T, the
The entire
magnet length is not shown for clarity, the total length being 3.32m.
required RF frequency is 298 MHz range, while a bore that
fits a human is at least 60 cm in diameter, making it above the
cutoff frequency for at least one mode of the bore viewed as a
II. EXPERIMENTAL SETUP
waveguide, when the waveguide is loaded with tissues which
The small-bore setup sketch in Fig.l shows the relevant
have high dielectric constants. The travelling waves can
dimensions and the position of the water phantom and patch
potentially be advantageous in terms of a more comfortable
probe. The total length of the 16.4 T superconducting magnet
environment for patients, larger field of view, imaging hard to
and bore is 332cm, not drawn to scale in Fig.l. The gradient
reach organs (e.g., prostate gland) and some areas
that were
coil part of the bore is smaller in diameter than the outer part,
until now inaccessible to MRI (e.g., inner ear), as well as
the diameters being 12cm and 26 cm, respectively. The larger
enabling new spatial encoding schemes and a variety of mode
cylindrical portion to the left where the probe is placed is
sensitivity profiles [8].
100cm long, and a similar 138-cm long portion is to the right
978-1-4799-8275-2/15/$31.00 ©2015 IEEE
of the gradient coil region. The inside wall of the bore is
insulated. The water phantom is 34cm long and fits snuggly in
the smaller cylindrical part of the bore.
The probe can be
translated through the bore, which results in a change in
Outer part
matching, as well as variations in the excitation field in the
ofbor�
phantom. It is fed by 3 kW of power at the Larmor frequency
of 698 MHz.
(a)
III. PROBE DESIGN
The probe designed for the 16.4-T bore from Fig.1 is a
circular
single-coaxial
feed
slotted
patch.
The
probe
is
designed using full-wave simulations (Ansoft HFSS), with the
bore and phantoms taken into account. The phantom is
simulated as a uniform dielectric cylinder with £,=81 and a
4
conductivity of 1O- S/m. The substrate is an FR4 62.5-mil
thick double-sided copper printed circuit board. The patch
feed is a single coaxial input as shown in Fig.2, and the two 5mm wide slots in the ground plane are 6.85 and 7.1cm long
and excite circularly-polarized waves [14].
(b)
Fig. 3.
(a) Photo of patch probe placed 50 cm into the bore. (b)
Simulated return loss of the patch probe when placed inside the bore
without a matching circuit (red) and measured return loss in the bore
�ith .an electrically short high-impedance line microstrip matching
cIrcuIt connected between the coaxial cable and patch feed point.
IV. 16.4-T MRI IMAGES
(a)
Fig. 2.
(b)
Photograph (a) and layout (b) of circularly-polarized patch
probe with a single coaxial feed and two slots in the ground plane
that ensure circular polarization.
The measured and simulated return loss, calibrated to the
coaxial feed reference plane, is shown in Fig.3. The patch
itself had about 8dB return loss, so a narrowband microstrip
matching circuit is designed to obtain a return loss greater than
20dB when the probe is placed in the bore.
Note that although some antenna theory and approaches are
used in the design of the patch, it is not actually an antenna in
free space, but rather a probe that couples to modes in the
loaded bore waveguide. Therefore, the simulations are done
inside a metal cylinder the size of the bore, and the distance to
the gradient coil portion is varied to verify the traveling wave
conditions. Coupling of the circularly-polarized field into the
phantom is also examined in simulations by observing the
transverse components of the B-field.
The phantom used for the experiments is a dielectric acrylic
tube (L=34 cm, D=9 cm) filled with deionized water or saline.
With the patch probe placed 50cm into the bore, as in Fig.3a,
MR images consistent with simulated B-field are obtained and
are shown in Figure 4.
In order to improve the SNR, A cylindrical array of 3-cm
wide
longitudinal
copper
strips 2-m
long
and 12cm
in
diameter is inserted in the bore with a goal of modifying the
travelling-wave mode content in the gradient coil region, as
shown in Figure 5. The longitudinal strips present an artificial
electromagnetic surface and modify the boundary conditions
of the metallic bore. Since they are oriented in the axial
direction, the strips present a different surface impedance to
the axial electric field than to the transverse field, thus
modifying the mode profile. This can be seen in both
simulations and measurements of the axial
cross-section
shown in Figure 6(a). The coronal images are in good
agreement with field simulations as shown in Figure 6(b). For
these images, the patch probe is placed Scm from the edge of
the bore at the start of the strip cylinder. The MR images were
obtained
with
a
GRE
sequence:
FOV=20x20
cm,
TRlTE=275ms/2.1ms, slice thickness of 1 mm for both axial
and coronal slices. The high spatial frequency fringes and
978-1-4799-8275-2/15/$31.00 ©2015 IEEE
edge brightening correspond to artifacts due to unencoded
and edge brightening correspond to artifacts due to unencoded
volume of the dielectric guide that extends beyond the
volume of the dielectric guide that extends beyond the
gradient insert.
gradient insert.
(a)
(b)
(a) Measured axial (top) and sagittal (bottom) cross-sections
Fig. 6. Magnitude of the simulated transverse right handed circular­
inside the phantom. (b) Normalized simulated circularly polarized B­
polarized H-Field normalized to 1W power input at the SMA
Fig. 4.
field for the sagittal and axial cuts, using COMSOL Multiphysics.
connector of the probe (left) and MR images (right) taken with the
patch probe and metal strip cylinder inserted in the bore, at a specific
In order to improve the SNR, A cylindrical array of 3-cm
wide
longitudinal
copper
strips 2-m
long
and 12cm
in
diameter is inserted in the bore with a goal of modifying the
cross-section of the phantom for (a) axial and (b) coronal cross­
sections. The SNR is increased seven-fold compared to the images in
Fig.4.
travelling-wave mode content in the gradient coil region, as
V. CONCLUSIONS AND DISCUSSION
shown in Figure S.
Copper strip cylind �r
I
The circular patch probe was chosen after an investigation
of other possible probes that result in similar magnetic field
profiles in the phantom. A rectangular patch probe was also
investigated with dimensions shown in Fig.7 and designed on
�
Patch probe
76cm
Gradient coil region
Fig. 5. Sketch of parallel copper strip cylinder inserted into the bore
the same type of substrate as the circular patch probe,
resulting in a return loss of 18dB when simulated inside the
bore.
and extending beyond the gradient coil region by 5cm. The patch
probe is placed at the beginning of the cylinder.
The longitudinal strips present an artificial electromagnetic
surface and modify the boundary conditions of the metallic
bore. Since they are oriented in the axial direction, the strips
�
7c
than to the transverse field, thus modifying the mode profile.
This can be seen in both simulations and measurements of the
axial cross-section shown in Figure 6(a). The coronal images
are in good agreement with field simulations as shown in
Figure 6(b). For these images, the patch probe is placed Scm
.9
3
present a different surface impedance to the axial electric field
�
10.2 cm
18 cm
m
�
from the edge of the bore at the start of the strip cylinder. The
MR images were obtained with a GRE sequence: FOV=20x20
cm, TRlTE=27Sms/2.lms, slice thickness of 1 mm for both
axial and coronal slices. The high spatial frequency fringes
Fig. 7. Layout of circularly-polarized rectangular patch with a single
coaxial feed in the lower right.
978-1-4799-8275-2/15/$31.00 ©2015 IEEE
u (T)
1.1\leiI -�n
I. 1.5010-009'
6. EtI�1Ie .. OOg
S.IIS I.-oet
S.5Il290-009
,. 9S!B._�
't,IIC'+S.c:-oe9
�. 1ISSIU-009'
1. �)c-ae!ii
ZO.7561te-oe1i
.2,2nl1l1!!;-OOG;
L l!.:S'Ne-OOg.
1.I�ee.-oeg
Fig. 8. Magnitude of the transverse right handed circular-polarized B­
field normalized to 1W power input at the SMA connector of the
probe for axial (top) and sagittal (bottom) simulated for the circular
patch (left) and the rectangular patch (right) in HFSS. These images
do not perfectly match the experimental image of Fig. 4 due to the
Fig. 9. GRE (a-c) and SE (d-t) images obtained in phantom with
patch and metal strip cylinder: (a,d) axial, (b,e) coronal, and (c,t)
sagittal slices. (SE parameters: TRlTE=ls/20ms, FOV=20x20 cm,
matrix=256xI28, slice thickness of 2 mm, 0. 8 mm gap, 10 slices).
MR image being a result of both polarizations.
Figure 8 shows a comparison of the simulated magnetic
field inside the phantom for the two probes. The simulations
are performed for similar conditions as in the experiment. The
simulations imply that the magnetic field profile is relatively
independent of the type of probe that is chosen, as waveguide
REFERENCES
[I]
[2]
disadvantage because of its size, and the reduced size of the
circular probe is a result of significant design efforts. In
206--218 (1994).
[3]
bore, and can also be scaled to fit inside 3T, 7T and lOT
static field MRI systems.
[4]
cylinder is inserted into the bore, modifying the modal
content. Although the guide diameter is too small to support
[5]
cylinder changes the boundary condition thus changing the
increase in
SNR as compared to the patch probe without the
were able to obtain spin-echo images, with an example shown
of
V. D. Schepkin, et al. 'In vivo Rodent Sodium and Proton MR
A. G. Webb, et af. "MRI and localized proton spectroscopy in
human leg muscle at 7 tesla using longitudinal traveling
[7]
K. Haines, et af. "Microimaging with a Cylindrical Ceramic
Dielectric Resonator at 21. 1 T," Proc. Exp. NMR Conf. 20IO.
[8]
A.lM. Kiruluta, "The emergence of the propagation wave
vector in high field NMR:
analysis and implications," J.
Physics D-Applied Physics, 40(10). 3043-3050. 2007.
[9]
F. H. Geschewski, et al., "Optimum coupling and multimode
excitation of traveling-waves in a whole-body 9.4T scanner,"
Magn. Reson. Med., vol. 69, no. 6, pp.1805-1812, June, 2013.
[10]
A. Tonyushkin, et ai, "Traveling Wave MRI in a Vertical Bore
[II]
A. Tonyushkin, et ai, "Imaging with Dielectric Waveguide
21.1-T System," Intern. Magn. Reson. Med., Melbourne, 2012.
in Fig.9. These results demonstrate that traveling wave MRI
Approach for 3T MRI,", Intern. Soc. Magn. Reson. Med.,
Melbourne, 2012
with high SNR can be performed with proper probe exciters
and bore design when the bore is small compared to the free­
laboratory.," Journal
waves," Magnetic Reson.Med., 63 (2010) 297-302.
seven-fold
strip cylinder. Due to the high SNR, in addition to GRE, we
R. Fu, et al. "Ultra-wide bore 900 mhz high-resolution nmr at
Imaging at 21 T," Magn. Reson. Imaging, 28(3). 400-7. 2010.
traveling waves above cut-off for a metal guide, the strip
mode cutoff frequencies, allowing TW MRI, with a
Brunner, et al. "Travelling-wave nuclear magnetic
Magnetic Resonance 177 (2005) 1-8.
[6]
phantom excited by a travelling-wave field patch probe far
O.
the national high magnetic field
In summary, this paper demonstrates 16.4T MR images in a
from the phantom. In addition, an anisotropic copper strip
D.
resonance," Nature 457(7232}. 994-U2. 2009.
contrast to the rectangular patch probe, the circular probe is
small enough and can fit in the gradient coil portion of the
IT. Vaughan, et al. "High frequency volume coils for clinical
NMR imaging and spectroscopy," Magn. Reson. Med. 32,
theory suggests. However, the location of the probe will
influence the field distribution. The rectangular probe has a
P.B. Roemer, et al. "The NMR phased array," Magn. Reson.
Med. 16, 192-225 (1990).
[12]
J.
A.
Tang,
et
al.,
"Cutoff-free traveling
wave NMR,"
Concepts in Magnetic Resonance Part A, vol. 38A, no. 5,
space wavelength of the Larmor frequency.
pp.253-267, September, 2011.
[13]
AJE Raaijmakers, et ai, "Design of a radiative surface coil
array element at 7T: The single-side adapted dipole antenna,"
Magn.Reson. Med., vol. 66, pp.1488-1497, November 2011.
VI. ACKNOWLEDGEMENTS
[14] X. L. Biao, M. l Ammann, lET Electronic Lett., Vol.42, No.4,
The authors acknowledge support by the National Science
Foundation
under
a
collaborative
research
grant
Feb. 2006.
ECCS
1307614 at the University of Colorado, Boulder.
978-1-4799-8275-2/15/$31.00 ©2015 IEEE