Shin-ichi Urayama1, Masaki Fukunaga2, and Martijn Cloos3
1Human Brain Research Center, Graduate School of Medicine, Kyoto University, Kyoto, Japan, 2National Institute for Physiological Sciences, Okazaki, Japan, 3University of Queensland, Brisbane, Australia
Synopsis
Keywords: RF Arrays & Systems, RF Arrays & Systems, RF-mode, non-pTx
To mitigate B1 inhomogeneity problems in
ultra-high-field MRI, we proposed RF-mode switching technique which doesn’t
need an expensive parallel transmit (pTx) system. As a preliminary study, we developed
an RF-switching circuit (one of the main parts of the system) and validated its
function through phantom experiments. The results
showed that the switching function works well, although a certain degree of
power-loss was observed both at bench (< -1dB) and on the images. We
have re-designed a new switching circuit and will continue to the development
to demonstrate this technique at a fraction of the cost of a pTx system.
Background
Although ultra-high field (UHF) MRI offers many
opportunities by virtue of its high SNR and strong contrast, B1 inhomogeneity
degrades image quality and can limit its usage in clinics and research. To
mitigate this problem, many studies have proposed [ex. 1,2]. In 2016, Cloos et al. [3] proposed that complementary
RF profiles, such as the circularly polarized (CP) and gradient modes, can be
interwoven in a fingerprinting scan to overcome B1 artifacts. Although this
idea was demonstrated using a pTx system, it should be possible to achieve the
same functionality with a much simpler and cheaper setup. The goal of our study
is to construct a dedicated Tx coil system with this RF-mode-switching function
on our non-pTx 7T scanner to demonstrate this technique at a fraction of the
cost of a pTx system. We introduce the basic concept and show preliminary experimental
results.Methods and Materials
Figure 1 shows
the basic design of the RF-mode-switching based transmit (Tx) system. It
consists of three devices connected in series: an RF-switching circuit (SC), a
Butler matrix [4], and an 8ch Tx coil. The Tx input to SC flows to either of the two
output ports depending on the TTL signal from the system cabinet whose timing
and duration are specified in our sequence code. And then, it will be divided into eight with hybrid couplers in the Butler matrix with phase shifting for CP
(if the current passes port-1) or gradient (port-2) mode to transfer into the
8ch Tx coil.
As a preliminary
study, we developed a prototype of SC (Fig.2a) and validated its function with
a 2ch Tx/Rx phased array coil. SC consists of three similar switch circuits
(one main and two sub-circuits) controlled with two bias inputs. Each circuit
has two quarter-wavelength coax cables [5]. When a bias
current is applied from the bias-1 (or 2) line, the Tx power from the main
circuits will flow through the right (or left) sub-circuit and out to the
Tx-output-1 (or 2). At the same time, the other coil is terminated with 50ohm
load. The S12 of SC was about -1 dB.
Validation
exams were carried out on a Magnetom 7T-MRI system (Siemens, Erlangen, Germany)
(Fig.2b). The Tx input port and one of the two bias lines of SC were connected
to an interface box which provided 1ch Tx and bias current from the patient
table connector and received Rx signals through SC (in the future TR switches
will be moved closer to toil past the SC). The two output ports of SC were hooked
up to the 2ch Tx/Rx phased-array coil with a bottle phantom filled with 0.5%
NaCl solution. Phantom images were obtained using a gradient-echo (GRE) sequence
with a small flip-angle (4V Tx voltage ) and compared with ones measured without SC between the interface
box and the coil. To calculate the SNR-map of each image, two GRE images were
obtained one with 0V Tx voltage and the other with 4V.Results
Figure 3 shows the SNR maps obtained from the two
coil elements with/without SC. Similar coil sensitivity patterns between maps
with and without SC indicate that SC effectively delivers the Tx current to only
one of the two channels of the coil and that the other coil is properly
terminated with a 50Ohm load (or else it would couple). However, more than twice the difference in the SNR between them suggests a
certain degree of power-loss existed in the present SC.Discussion
Despite plenty of studies aiming to
solve the B1 inhomogeneity problem in UHF-MRI, only adiabatic pulses are used
universally available. Because the RF-mode switching technique can be made
affordably and doesn’t need a pTx system, it could be accessible to wider audiences.
The RF switch
circuit we produced in this preliminary study showed the switching function
works well, although we observed
slightly more power-loss than expected.
Probably this loss was caused by insufficient tuning of the circuits, cheap
components on this test board, and redundancy in the design as a prototype. Since the received MR signal also pass-through SC, the observed loss
in SNR is a combination form losses in transmit power and added noise in the
receive chain. To reduce the transmit losses, we have designed a compact and
more efficient switching circuit on a single PCB (Fig. 4) and will produce it
at the next step. In addition, we will move the preamplifiers closer to the
coil behind the SC.
Except
for SC, there are several devices we will have to develop for an RF-mode
switching system with optimal performance, such as a bias current switching
circuit, the Butler matrix, an 8ch Tx coil, and an Rx-only phased-array coil. The
bias current switching circuit is a part of SC that provides the current to one
of the two bias inputs of SC based on the TTL signal from the cabinet.
[MC1]And cheap components on this test board.Conclusion
To mitigate the B1 inhomogeneity problem at non-pTx
UHF-MRI, we proposed the RF-mode switching system. Here, we built an RF
switching circuit as a preliminary study and
demonstrated the usefulness of the circuit.Acknowledgements
This work was supported by JSPS KAKENHI (Grant Nos. JP19H03601).References
1.
Green EM, B1 insensitive
techniques for ultra-high field magnetic resonance imaging, PhD Thesis, http://hdl.handle.net/11343/311718
(2021)
2.
Ladd
ME, et al., Pros and cons of ultra-high-field MRI/MRS for human application.,
Prog Nucl Magn Reson Spectrosc 109:1–50 (2018)
3.
Cloos
MA, et al., Multiparametric imaging with heterogeneous radiofrequency fields. Nat
Commun 7:12445. (2016)
4.
Alagappan
V, et al., Degenerate Mode Band-Pass Birdcage Coil for Accelerated Parallel
Excitation, Magn. Reson. Med., 57:1148–1158 (2007)
5.
Ji
Y, et al., High peak and high average radiofrequency power transmit/ receive
switch for thermal magnetic resonance, Magn. Reson. Med., 80:2246–2255 (2018)