Max van Riel1,2, Zidan Yu2,3, Shota Hodono2,3, Ding Xia2, Hersh Chandarana2, and Martijn A. Cloos2,3
1Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands, 2Center for Advanced Imaging Innovation and Research and Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York University School of Medicine, New York, NY, United States, 3Sackler Institute of Graduate Biomedical Sciences, New York University School of Medicine, New York, NY, United States
Synopsis
We demonstrate a
3D MR fingerprinting sequence for quantitative abdominal imaging. The sequence
was made robust to motion by modifying the order of acquisition to allow for
free-breathing imaging. Furthermore, the flip angle pattern was optimized using
the Cramér-Rao Lower Bound to increase the efficiency of the sequence. A
phantom was used to validate the new sequence and it was shown that the motion-robust
ordering reduced motion-related artefacts. In vivo results showed reduced
artefacts as well. We conclude that it is possible to generate B1-robust quantitative T1 and proton density maps at a clinically usable
resolution within 5 minutes.
Introduction
Magnetic Resonance Fingerprinting (MRF) is a technique that allows
for fast, simultaneous quantification of multiple parameters, such as T1, T2,
proton density (PD), and/or B1 field strength.1,2 Extending MRF to 3D allows
for a larger coverage and a better through-plane resolution.3,4 However, when imaging the
abdomen, motion can reduce the quality of the parameter maps.5 Moreover, B0 and
B1 inhomogeneities are particularly problematic, due to air pockets
and destructive interference of the B1 field.6 In this work, we present a
motion-robust 3D MRF sequence for quantitative abdominal imaging. The method
was validated using a moving phantom and applied in vivo during free-breathing
measurements. Methods
Sequence design
To generate
the fingerprints, a Fast Low-Angle SHot (FLASH) sequence with variable flip
angles was used. All excitations were RF- and gradient-spoiled. A non-selective
adiabatic inversion pulse was applied before the start of the sequence to
enhance the T1-encoding capability of the fingerprints.
The k-space data was acquired in a 3D stack-of-stars trajectory.7 Originally, with every
subsequent excitation, the spoke with the same phase encoding index but a
different azimuthal angle was acquired (normal
ordering). In the proposed free-breathing sequence, the kz-index
of the acquired line is increased by one with every excitation while keeping
the azimuthal angle constant (motion-robust
ordering, see Figure 1). This way, adjacent spokes
in the phase encoding direction are acquired in quick succession, reducing motion-related
artefacts. This creates several sets, where each set consists of as many
consecutive spokes as the number of slices.
To minimize
the variance of the estimated parameters, the flip angle pattern was optimized using
the relative Cramér-Rao Bound (CRB)8–12 for T1 and B1. The peak flip angle was
limited to 60 degrees to ensure that the necessary transmit voltage could be
implemented on the scanner. Sequences with 300, 600, 900, and 1800 flip angles
were optimized (Figure 2 a-d) in MATLAB (The MathWorks Inc., Natick, MA,
USA), using the CasADI13
toolbox for automatic differentiation and the Interior Point OPTimizer (Ipopt)14 as
optimization algorithm.
Image acquisition
All experiments
were performed on a clinical 3T MRI scanner (Prisma, Siemens, Erlangen, Germany).
A phantom containing glass tubes with different T1 values was placed on a cart
made from LEGO® (The Lego Group, Billund, Denmark), riding on a slope of ~7
degrees and controlled by a motor outside the scanner room (RWTH Mindstorms NXT
Toolbox for MATLAB, RWTH Aachen University, Germany). An 18-channel body coil
was placed over the phantom assembly during the scan (Figure 2). The same coil
was used for the in vivo scan. The parameters used for all imaging experiments are
summarized in Table 1. The
acquisition time was kept constant by acquiring more shots for shorter
sequences. A previously published 2D MRF implementation15 was used as a reference scan to validate the quantitative T1 values
without motion.
Dictionary construction
Our dictionary contained 17600 fingerprints, each with a unique
combination of T1 and B1 values. The T1 values used for the
dictionary ranged from 50 ms to 3764 ms with increments of 2.5%, while the relative
B1 field strengths ranged from 0.02 to 2.0 in steps of 0.02. The
Bloch equations were used to simulate the effect of the RF-pulses on a single
isochromat for each parameter combination. The resulting signals within each
set were averaged and each fingerprint was normalized to have unit Euclidean
norm.
Dictionary matching
Each set was
reconstructed separately using the Nonuniform Fast Fourier Transform (NUFFT)16, resulting in a sequence of images. The best matching dictionary
entries for every voxel were then used to obtain the parametric maps.1Results & Discussion
The results from
the phantom scan can be seen in Figure 4. The
sequence with 300 flip angles was not as accurate for higher T1 values. Most
likely, the time from one inversion to the next is too short to observe the
slow dynamics of the long T1 samples. The longest sequence (with the fewest
number of shots) showed too much variability in the T1 estimates due to the
increased undersampling artefacts. The sequences with 600 and 900 flip angles demonstrated
fewer artefacts and better agreement with the reference when using the motion-robust
ordering.
The motion
artefacts were not visible for all periods of the motion pattern when not using
the motion-robust ordering. This was probably caused by interference with the
timing of the sequence due to the regular motion of the phantom. Since in vivo
breathing motion is not as regular and may occur at any frequency, the proposed
ordering is important to prevent motion artefacts.
The sequence
with 600 flip angles was selected to be tested in vivo. The motion-robust
ordering showed better detail in the T1 map and reduced the motion-related
artefacts (Figure
5). In
particular, a clear artifact is visible in both the T1 and B1 maps
acquired using the normal ordering.Conclusion
A free-breathing
MR Fingerprinting sequence was demonstrated for B1-robust
quantitative abdominal imaging. The sequence was validated in a phantom and
demonstrated in vivo. With this motion-robust MRF implementation it is possible
to collect crisp PD images and accurate T1 maps of the abdomen at a clinically
usable resolution within 5 minutes.Acknowledgements
The research
reported in this publication was supported by the NIH/NIBIB grant R01 EB026456,
NIH/NIAMS grant R01 AR070297, and performed under the rubric of the Center for
Advanced Imaging Innovation and Research, an NIBIB Biomedical Technology
Resource Center (P41 EB017183). Furthermore, I would like to express my
gratitude towards the Holland Scholarship from the Dutch Ministry of Education,
Culture and Science and the Amandus H. Lundqvist Scholarship Program for supporting
this internship project.References
1. Ma D, Gulani
V, Seiberlich N, et al. Magnetic resonance fingerprinting. Nature.
2013;495(7440):187-192. doi:10.1038/nature11971
2. Cloos MA,
Knoll F, Zhao T, et al. Multiparametric imaging with heterogeneous
radiofrequency fields. Nat Commun. 2016;7:1-10. doi:10.1038/ncomms12445
3. Ma D, Jiang
Y, Chen Y, et al. Fast 3D Magnetic Resonance Fingerprinting for a Whole-Brain
Coverage. Magn Reson Med. 2018;79(4):2190-2197. doi:10.1002/mrm.26886
4. Fujimoto K,
Cloos MA, Urushibata Y, Okada T. Cortical T1 mapping with 3D MR Fingerprinting
at 7T using a single transmit channel. In: Proceedings of the 26th Annual
Meeting of ISMRM, Paris, France; 2018:1135.
5. Yu Z, Zhao
T, Assländer J, Lattanzi R, Sodickson DK, Cloos MA. Exploring the sensitivity
of magnetic resonance fingerprinting to motion. Magn Reson Imaging.
2018;54:241-248. doi:10.1016/j.mri.2018.09.002
6. Merkle EM,
Dale BM. Abdominal MRI at 3.0 T: The basic revisited. Am J Roentgenol.
2006;186(6):1524-1532. doi:10.2214/AJR.05.0932
7. Block KT,
Chandarana H, Milla S, et al. Towards Routine Clinical Use of Radial
Stack-of-Stars 3D Gradient-Echo Sequences for Reducing Motion Sensitivity. J
Korean Soc Magn Reson Med. 2014;18(2):87-106.
doi:10.13104/jksmrm.2014.18.2.
8. Kay SM. Fundamentals
of Statistical Signal Processing: Estimation Theory. Vol 1. Upper Saddle
River, NJ: Prentice Hall PTR; 1993.
9. Teixeira
RPAG, Malik SJ, Hajnal J V. Joint System Relaxometry (JSR) and Cramér-Rao Lower
Bound Optimization of Sequence Parameters: A Framework for Enhanced Precision
of DESPOT T1 and T2 Estimation. Magn Reson Med. 2018;79(1):234-245.
doi:10.1002/mrm.26670
10. Zhao B, Haldar
JP, Liao C, et al. Optimal Experiment Design for Magnetic Resonance
Fingerprinting: Cramér-Rao Bound Meets Spin Dynamics. IEEE Trans Med Imaging.
2019;38(3):844-861. doi:10.1109/TMI.2018.2873704
11. Lee PK,
Watkins LE, Anderson TI, Buonincontri G, Hargreaves BA. Flexible and efficient
optimization of quantitative sequences using automatic differentiation of Bloch
simulations. Magn Reson Med. 2019;82(4):1438-1451. doi:10.1002/mrm.27832
12. Assländer J,
Lattanzi R, Sodickson DK, Cloos MA. Optimized quantification of spin relaxation
times in the hybrid state. Magn Reson Med. 2019;82(4):1385-1397.
doi:10.1002/mrm.27819
13. Andersson JAE,
Gillis J, Horn G, Rawlings JB, Diehl M. CasADi – A software framework for
nonlinear optimization and optimal control. Math Program Comput.
2019;11(1):1-36. doi:10.1007/s12532-018-0139-4
14. Andreas W,
Biegler LT. On the implementation of an interior-point filter line-search
algorithm for large-scale nonlinear programming. Math Program.
2006;106(1):25-57. doi:10.1007/s10107-004-0559-y
15. Cloos MA,
Assländer J, Abbas B, et al. Rapid Radial T 1 and T 2 Mapping of the Hip
Articular Cartilage With Magnetic Resonance Fingerprinting. J Magn Reson
Imaging. 2019;50(3):810-815. doi:10.1002/jmri.26615
16. Fessler JA,
Sutton BP. Nonuniform Fast Fourier Transforms Using Min-Max Interpolation. IEEE
Trans Signal Process. 2003;51(2):560-574. doi:10.1109/TSP.2002.807005