Ekin Karasan1, Victor Taracila2, Fraser Robb2, and Michael Lustig1
1Department of Electrical Engineering, University of California, Berkeley, CA, United States, 2GE Healthcare, Coils, Aurora, OH, United States
Synopsis
“In the light of the
moon a little egg lays on a leaf”[7]
Management of cabling connecting coil arrays to system remains the
nightmare of RF-engineers. Coupling within arrays to cabling reduces
performance, and coupling to transmit field causes high shield currents. Consequently,
traps are placed on conductors. To provide high blocking, traps are rigid,
large and often heavy, hindering flexibility. Instead of a few high blocking
traps, we propose a distributed system of small, elastic traps. We
leverage self-shielded resonant toroids forming a caterpillar-like structure.
We show that this design can attenuate shield currents while being robust to
flexing.
Introduction
“On Sunday.. out of the egg comes a tiny and very hungry
caterpillar”[7]
Common mode currents occur on conducting wires within the
bore of the scanner, for example, on the shield of coaxial cables between coils
and main system. These currents effect coil performance, result in B1 inhomogeneities
and pose serious safety hazards such as RF burns[1]. To mitigate shield currents,
resonant cable traps are placed at certain intervals along cable (Figure 1a). While
many designs exist[2], the basic principle is similar -- creating a high
impedance through a resonance LC tank.
These can either leverage the conductor itself, or inductively couple,
i.e. a “floating” trap[3]. Floating traps can be easily fitted and positioned
but don’t allow cable to be flexed and are only effective if placed in regions
of highest shield current. These structures couple, thus, cannot be place
closely, and need to have a length larger than radius in order to not interact
with the B1 field and perform efficiently[4]. For this reason, a few large,
high performance structures are used. Here, we propose a new, distributed
design of traps where smaller self-shielded substructures are placed along the full
length of cable (Figure 1b). These self-shielded structures do not couple to main
field or each other thus requiring no shielding. With this distributed design, any
bending is unlikely to impact overall structure allowing the cable to still be
flexible while providing sufficient attenuation (Figure 1c).Methods
The substructures were chosen to be resonant toroids as they
are mostly self-shielded, especially with increased number of windings. Toroids
have previously been used in traps but as the inductor for large, inductively
coupling structures[5]. Our smaller traps can be laid out, along the entirety of
the cable in a caterpillar-like structure without impeding its flexibility. The
traps are tuned using four capacitors, one of which is variable, so that electrical
field due to capacitors is uniformly distributed around structure.
Simulations
Initially, simulations were conducted with air-core toroids
to determine optimal parameters of the overall structure for a fixed wire and
toroid thickness($$$t$$$) including diameter
of toroids($$$d$$$), separation between
toroids($$$s$$$) and number of windings in
each toroid($$$N$$$) as shown in Figure 2. Simulation results normalized by total
length($$$l$$$) of the trap show that a
24-turn toroid outperforms a 12-turn toroid while slight coupling is observed between
toroids with increased toroid diameter as seen from the reduced performance of
the 12mm diameter toroid with $$$s=2$$$mm compared to $$$s=5$$$mm. From simulations, optimal
parameters were determined to be $$$d=9$$$mm, $$$s=5$$$mm and $$$N=24$$$ and the structure built
accordingly.
Fabrication
The design of a substructure is shown in Figure 3a. To allow
overall flexibility, the structure is manufactured on a low-loss
Polydimethylsiloxan(PDMS) elastic substrate. A mold is designed (Figure 3b)
with a slit for the wire and slots for the four capacitors. The mold is 3D
printed and used for injection molding. Toroids are hand wound with magnet wire
and the capacitors soldered on resulting in an elastic structure (Figure 3c).
Benchtop
Tests were conducted with setup in Figure 4a. Toroids were
fitted on a 6.80mm diameter cable. The performance of the structure with
different numbers of toroids was measured and compared to simulations and
theoretical ideal calculations, that assume low loss wires, vacuum medium, and
no interactions between toroids.
Scans
B1 distortion experiments were performed on an GE MR750WGE
(Waukesha, Wisconsin). Two cables were cut to the length of a resonant dipole
for 127.7MHz. Three B1-mapping measurements[6], with a flip angle of 30o,
were performed in a non-loading phantom. The first mapped the B1 field produced
by the body transmitter alone. In the second, the resonant dipole was placed on
the phantom, while measuring the B1 field in a slice 1.65 cm below wire. In the
3rd, a resonant dipole equipped with our caterpillar traps was placed in the
same way and measured. To evaluate the performance while flexing the wire, we
repeated the measurements, this time with a flip angle of 15o, by
bending wires and placing them on the phantom as shown in Figure 5.
Results and Discussion
A comparison between ideal, simulations and benchtop
measurements show that both simulations and benchtop measurements were inferior
to the ideal providing 3.64dB and 4.96dB reduced attenuation for single toroid
respectively (Figure 4b). The benchtop measurements are also slightly inferior
to simulations due to decreased permeability of PDMS compared to air. Benchtop
measurements show that a single toroid is able to obtain -5.11dB attenuation at
a resonant frequency of 127.7 MHz attaining a Q-factor of 106, whereas a 7
toroid structure is able to obtain -22.82dB attenuation attaining a Q-factor exceeding 350 (Figure 4c).
Figure 5 displays the results of the B1 distortion
experiments. The B1 field maps are displayed, in degrees. Attenuation of the
shield current by caterpillar traps are evident in both configurations.Conclusions
“Now he isn’t a little caterpillar. He is a big fat
caterpillar”[7]
Our distributed design splits the cable trap into smaller, self-shielded
subcomponents. This study shows that our caterpillar traps can provide
sufficient attenuation to shield currents while also maintaining flexibility,
enabling the entire structure from coil to system to be light, flexible and convenient. Acknowledgements
No acknowledgement found.References
- Beck BL, Peterson DM, Duensing GR, Fitzsimmons JR. 2000.
Implications of Cable Shield Currents at 3.0 and 4.7 Tesla. Int Soc Magn Reson
Med, Denver, CO. p 641.
- X. Yang and F. Fujita,
“T/R switches, baluns, and detuning elements in MRI RF Coils,” Proc. Intl. Soc.
Mag. Reson. Med. 14, Seattle, 2006.
-
Burl, M., Chmielewski, T., & Braum, W. O.
(2001). Patent No. US6593744B2.
- Seeber,
D. A., J. Jevtic, and A. Menon. "Floating shield current suppression
trap." Concepts in Magnetic Resonance Part B: Magnetic
Resonance Engineering: An Educational Journal
(2004): 26-31.
- Taracila V., Smith L.,
Stack C., Robb F. “Optimal Bazooka Balun Proportions,” Proc. Intl. Soc. Mag.
Reson. Med. 27, Paris, 2018.
-
Sacolick, Laura I et al. “B1
mapping by Bloch-Siegert shift.” Magnetic resonance in medicine vol. 63,5 (2010): 1315-22.
- Eric Carle “The Very
Hungry Caterpillar” 1969, ISBN-10 0399226907