Michael Garwood1, Michael Mullen1, Naoharu Kobayashi1, Lance delaBarre1, Steven Suddarth1, Djaudat Idiyatullin1, John Strupp1, Gregor Adriany1, Jarvis Haupt1, Alex Gutierrez1, Taylor Froelich1, Russell Lagore1, Benjamin Parkinson2, Konstantinos Bouloukakis2, Mark Hunter2, Mathieu Szmigiel2, Mailin Lemke2, Edgar Rodriguez-Ramirez2, Robin de Graaf3, Chathura Kumaragamage3, Scott McIntyre3, Terry Nixon3, Christoph Juchem4, Sebastian Theilenberg4, Yun Shang4, Jalal Ghazouani4, Alberto Tannús5, Mateus José Martins5, Edson Vidoto5, Fernando Paiva5, Daniel Pizetta5, Maurício Falvo5, Diego Turibio5, Christian Bones5, Eduardo Falvo5, John Thomas Vaughan4, Julie Kabil4, Hazal Yüksel4, Harish Krishnaswamy4, Sung-Min Sohn6, and Ramon Gilberto Gonzalez7
1Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, 2Victoria University of Wellington, Wellington, New Zealand, 3Yale University, New Haven, CT, United States, 4Columbia University, New York, NY, United States, 5Centro de Imagens e Espectroscopia por Ressonância Magnética, Universidade de São Paulo in São Carlos, São Carlos, Brazil, 6Arizona State University, Tempe, AZ, United States, 7MGH/Harvard, Boston, MA, United States
Synopsis
A multi-disciplinary
team of researchers in a multi-institutional consortium have designed and are
building an easily relocatable head-only 1.5T MRI scanner weighing only ~500
kg. The goal is to develop a radically new type of MRI scanner that will
enhance brain research, and ultimately, enable the diagnosis of neurological
diseases in underserved populations throughout the world where MRI scanners are
currently unavailable. To image with this system, pulse sequences have been developed and
implemented to generate images using a highly inhomogeneous B0.
Introduction
MRI scanning of people living in remote locations, and the ability to study certain human behaviors, are largely inaccessible by current MRI technology. For example, any behavior involving motion, and especially
those involving the upright real-time interaction with objects in natural
environments, cannot be studied. Such studies are of enormous scientific
interest, for example, in understanding the neuronal basis of motor planning,
but also of considerable practical and clinical importance in order to
eventually understand and address the motor deficits associated with injury,
stroke, or disease which preclude everyday behaviors as important as feeding
and reaching. In this project, we are building a compact head-only1.5T MRI scanner requiring only the head to be in the magnet bore, with a window for viewing outside the bore. Hardware Components of the Scanner
A CAD drawing in Fig. 1 shows the magnet, its
stand, and chair. This second-generation high temperature superconductor (HTS) magnet
is designed to allow the subject’s shoulders to be outside the magnet, with a
window for viewing outside the bore. The magnet is cryogen free, operating at 30
K using a single stage pulse-tube cryocooler. The magnet has been designed to
cool down rapidly (1-2 days), consuming only 5 kW power during steady state
operation. A water-cooled multi-coil
(MC) array consisting of 31 DC coils (Fig. 2) is used for dynamic shimming and
for producing spatial-encoding fields for imaging (1,2). Each channel of the MC array can be driven
independently and dynamically using a homebuilt multi-channel digital driver hardware
(3) and commercial amplifiers with +/-5 A (Resonance Research Inc)
for concurrent B0 shimming and linear or non-linear imaging. When configured to produce gradient fields, the
MC array achieves a maximum of 10 kHz/cm linear field gradient in any of the three
orthogonal directions. In addition, parabolic and hyperbolic field
shapes up to 500 Hz/cm2 can be produced for enabling non-linear spatiotemporal
encoding with methods such as STEREO (4). The
magnet cryogenics have been optimized to minimize eddy current coupling between
the MC array and the magnet. The digital spectrometer of this MRI system was specifically
designed for advanced research capabilities, and includes eight transmit/eight
receive RF channels and an additional field lock channel. The RF headcoil is tapered to conform to the tapered magnet
bore and has a modular design allowing operation in either circularly-polarized or 8-channel parallel transmission mode.
Software and Pulse Sequences of the Scanner
The
drastically decreased size of the magnet comes at the cost of greatly
diminished B0 uniformity, which after
passive shimming is predicted to be ~300 ppm over the adult human brain. Obviously,
narrowband pulse sequences like EPI will fail with such an inhomogeneous B0.
Thus, for this project to be successful, methods must be identified and/or
developed to permit high quality
MR imaging using a highly inhomogeneous B0. To accomplish this, only certain classes of pulse sequences
are feasible, and in some cases, new ones must be developed to produce the main
types of contrast (T1, T2, T2*, diffusion) needed for the proposed research and
clinical uses. In addition, the RF pulse and acquisition bandwidths in these pulse
sequences must be maximized, within SAR and SNR limitations, to fully excite
the brain and to minimize image distortion. Several sequences have been
developed and implemented to image with a highly inhomogeneous B0, including short-TE gradient-recalled echo (5) and missing-pulse steady state free precession (MP-SSFP) (6).
The user interface developed for this new MRI system consists of
software implemented with PyMR, a framework for programming MR
systems, which includes an Integrated
Development Environment (IDE) to help MR methods designers create
sequences with full graphical and unique programming capabilities (7, 8,
9). All that software is part of the operator console that allows
high-level post-processing of acquired datasets. Using these
features, sequence
designers can easily implement the most sophisticated pulse sequences
using some of the most modern tools available to software projects.Discussion
There is an urgent need for brain imaging
technology that is more portable and less restricting than current MRI
scanners. One way to address these issues is to decrease the size of the magnet to make a head-only system which does not confine the body; however, this
approach leads to drastically reduced B0 homogeneity
which, with current technologies, precludes high resolution imaging. In this project, we have addressed this
problem by developing new hardware, as well as new acquisition and
reconstruction methods, capable of producing high quality brain images despite
extreme B0 inhomogeneity. The goal is to demonstrate the first-ever high-field (1.5T) MRI scanner requiring only the head to be inside the magnet bore and having a
large window for viewing outside the magnet bore. The small size, weight, and
power requirements of this MRI system will enable it to be
transported and sited almost anywhere in the world and the MRI system can be brought to the subject rather than the other way around. By midyear 2020, system integration will begin, and
by late 2020, we hope to acquire the first images.Acknowledgements
This
work was supported by the National Institutes of Health grants U01 EB025153 and
P41 EB015894.References
1) Rudrapatna
SU, Fluerenbrock F, Nixon WT, de Graaf RA, Juchem C. Combined imaging and
shimming with the dynamic multi-coil technique. Magn Reson Med;81(2):1424-1433, 2019
2) Juchem
C, Mullen M, Kumaragamage C, DelaBarre L, Adriany G, Brown PB, McIntyre S,
Nixon TW, Garwood M, de Graaf RA. Dynamic Multi-coil technique (DYNAMITE) MRI
on human brain. 2019; Montreal, QC, Canada. p 0219
3)
Nixon
TW, McIntyre S, de Graaf RA. The design and implementation of a 64 channel
arbitrary gradient waveform controller. 2017; Honolulu, HI, USA. p 0969
4)
Snyder ALS, Corum CA, Moeller S, Powell NJ, Garwood M, MRI by steering resonance through space. Magn Reson Med 72:49–58, 2014
5) Mullen M, Kobayashi N, Garwood M. Two-dimensional Frequency-swept pulse with
resilience to both B1 and B0 inhomogeneity. J Magn Reson 299:93-100,
2019
6) Kobayashi N, Idiyatullin D, Adriany G, Garwood, M. Magnetic resonance imaging under highly inhomogeneous B0 fields using missing-pulse steady-state free precession (MP-SSFP). 2017; Honolulu, HI, USA. p 5052
7) Pizetta DC, Lourenço GV, Silva DMDD, Vidoto ELG,
Martins MJ, Tannús A 2015, Magnetic resonance system configuration and
editing tools, Jaffray DA (Ed.), IUPESM 2015: Wold congress on medical
physics & biomedical engineering, Toronto, Canada, Jun 7-12,
p667-668.
8) Pizetta DC,
Shimada DHD, Falvo M, Souza PVBD, Silva LR, Bittencourt HP, Vidoto ELG,
Martins MJ, Tannús A, Universidade de São Paulo 2017, PyMR - a framework for programming magnetic resonance systems, BR-512019001829-0.
9) Pizetta
DC 2018. PyMR: a framework for programming magnetic resonance
systems, PhD Thesis, Universidade de São Paulo, São Carlos,
doi:10.11606/T.76.2019.tde-06052019-103714.