Hampus Olsson1, Mads Andersen2, Mustafa Kadhim1, and Gunther Helms1
1Department of Medical Radiation Physics, Clinical Sciences Lund, Lund University, Lund, Sweden, 2Philips Healthcare, Copenhagen, Denmark
Synopsis
MP2RAGE has become popular for T1 mapping
at 7T. Accuracy is yet improved by using a separately acquired flip angle map
when creating the protocol-specific lookup table. Here, two additional volumes
acquired at differing flip angles are added to an MP2RAGE sequence to obtain
two separate states of driven equilibriums, effectively forming a dual flip
angle protocol within the cycle. By estimating the accelerated effective relaxation,
T1*, from the signals, both T1 and B1+
can be solved for analytically. Thus, a multi-volume MPRAGE sequence is turned
into a dedicated high-resolution T1 and flip angle-mapping protocol.
Introduction
The well-known MP2RAGE sequence can be used to acquire
maps of T1 by a protocol-specific lookup table (LUT) of the signal
evolution1. Through a separately
acquired flip angle map, the residual influence of transmit B1+
inhomogeneities can be alleviated. The concept has been expanded upon through acquisition of a third volume(dubbed “MP3RAGE”). Either, to correct for spatially varying
inversion efficiency2 or to acquire a
map of B1+ in addition to T13. The latter approach ideally
requires that one pair of acquisitions is only sensitive to T1 while
the other is only sensitive to B1+.
Here we propose an alternative to the LUT approach using a driven equilibrium (DE) MPRAGE sequence with ≥3 volumes.
First, two volumes ($$$S_1$$$, $$$S_2$$$) are acquired
using small flip angles during relaxation with time constant T1* to
a PD-weighted DE (Figure 1). Secondly, two additional
T1-weighted volumes ($$$S_3$$$, $$$S_4$$$) are
acquired directly at a lower DE by a higher flip angle. These two DEs form a dual flip
angle T1-mapping scheme with a quadratic dependence on B1+4. This, in combination with
fitting of the signal equation to obtain T1*, allows to analytically
solve for both T1 and B1+, at the high resolution
provided by MPRAGE.Methods
The time constant, $$$T_1^*$$$, and the corresponding PD-weighted DE, $$$S_{PD}$$$, which the longitudinal magnetization relaxes towards was derived through fitting $$S_i=S_{PD}+(S_0-S_{PD})\exp(-TI_i/T_1^*)$$ where $$$S_0\approx-f_{inv}S_4\sin(\alpha_{PD})/\sin(\alpha_{T1})$$$ with $$$f_{inv}=0.96$$$ as in ref1. In addition, $$$S_{PD}$$$ was also obtained from a separate
acquisition with the same $$$TR$$$ and flip angle to validate the
concept. The apparent $$$T_{1,app}=T_1/f_T^2$$$, where $$$f_T$$$ is the B1+
bias, was calculated as4 $$T_{1,app}=2TR\frac{S_{PD}/\alpha_{PD}+S_4/\alpha_{T1}}{S_4\alpha{T1}-S_{PD}\alpha_{T1}}.$$ Using
small flip angle approximations as above gives $$T_1^*=(T_1^{-1}+\frac{f_T^2\alpha_{PD}^2}{2TR})^{-1}$$ where $$$f_T^2$$$ can be replaced by $$$T_{1,app}$$$ and then $$$T_1$$$ solved for as: $$T_1=T_1^*(1+\frac{\alpha_{PD}^2T_{1,app}}{2TR}).$$ Finally, the B1+ map was obtained as: $$f_T=\sqrt{(T_{1,app}/T_1)}.$$
A healthy
adult subject (male, 29 years old) was scanned on a 7T Philips Achieva MR
system (Philips Healthcare, Best, NL), using a dual channel transmit head coil
with 32 receive elements (Nova Medical, Wilmington, MA) after giving informed
written consent.
Four interleaved volumes of 1 mm isotropic resolution were acquired without delays at TI=1012/3009/5006/7003 ms and α=3°,3°,16°,16° (αPD,αT1=3°,16°), TR/TE=7.8/3.4 ms and
TF=256, resulting in a cycle duration of TC=8004 ms. The duration of all
readout pulses were 700 μs to balance incidental magnetization
transfer effects5. Application of SENSE=2, PF=0.775 and an
elliptical k-space readout resulted in a total acquisition time of 7:12 min.
As a reference for B1+,
four DREAM6 maps with preparation flip angles of 25°,40°,60°,90° at 3.75x3.75x3.50 mm3 were
acquired and combined for better accuracy7.
A LUT-based approach was also performed where the
signal evolution for all combinations of 500≤T1≤4500 ms and 0.4≤fT≤1.6
was simulated and compared to the measured signal in $$$S_{1,2,3,4}$$$ normalized to $$$S_1$$$.Results
Without an external reference for $$$S_{PD}$$$,
there was a positive bias in the derived T1 map which translated
into a T1-related bias in the B1+ map (Figure
2). With the LUT-based approach, a global minimum for a specific T1 and B1+ was obtained but mirrored this overestimation of T1 (Figure 3). With an independent PD-weighted reference, a spatially homogenous T1 map that agreed with literature estimates8 was obtained although high
T1 pixels such as in the CSF and to a lesser extent cortical gray
matter appeared somewhat underestimated (Figure
4). The corresponding B1+
map showed excellent agreement with the lower resolution DREAM reference method
(Figure
5) although in low B1+
regions such as the cerebellum, a somewhat lower B1+ was
obtained.Discussion
The DE-MPRAGE-derived T1 map looks
promising and the B1+ map showed excellent agreement with
the reference DREAM method, effectively validating the proposed method. However, protocol optimization is needed to make the approach independent of a separate $$$S_{PD}$$$ measurement. The simulation in Figure 3 shows that the problem is
well-defined. By reading out the center of k-space at the end of each acquisition train (reverse central encoding), the approach could be substantially improved. This would improve the definition of both DE states, and thus elimate the need for a 4th volume (only needed to obtain the DE in low B1+
areas), thus reducing acquisition time by 25%. Conclusion
We show that it is possible to obtain high resolution
3D maps of T1 and B1+ at 7T using a DE-MPRAGE
sequence with multiple volumes. Thus, effectively expanding MP2RAGE to a
dedicated T1 and B1+-mapping sequence while
foregoing the use of a lookup table.Acknowledgements
Grant support by the Swedish Research Council (NT 2014-6193). Lund University Bioimaging Center (LBIC) is acknowledged for experimental resources (equipment grant VR RFI 829-2010-5928).References
1. Marques JP, Kober T, Krueger G, van der Zwaag W, Van de
Moortele PF, Gruetter R. MP2RAGE, a self bias-field corrected sequence for
improved segmentation and T1-mapping at high field. Neuroimage 2010;49(2):1271-1281.
2. Hung W, Chen P, Chuang T, Chang H, Wu M. High resolution volumetric T1 mapping using a novel MP3RAGE method. In Proceedings of the 21st Annual Meeting of ISMRM, Salt Lake City, Utah, USA, 2013. p. 2353.
3. Rioux
JA, Saranathan M, Rutt BK. Simultaneous Whole-Brain T1 and Flip Angle
Mapping with MP3RAGE. In Proceedings of the 22nd Annual Meeting of
ISMRM, Milan, Italy, 2014. p. 3216.
4. Helms G, Dathe H, Dechent P.
Quantitative FLASH MRI at 3T using a rational approximation of the Ernst equation.
Magn Reson Med 2008;59(3):667-672.
5. Olsson H, Andersen M, Latt J, Wirestam
R, Helms G. Reducing bias in dual flip angle T1 -mapping in human brain at 7T.
Magn Reson Med 2020;84(3):1347-1358.
6. Nehrke K, Bornert P. DREAM-a novel
approach for robust, ultrafast, multislice B1 mapping. Magn Reson
Med 2012;68(5):1517-1526.
7. Olsson H, Andersen M, Helms G. Reducing
bias in DREAM flip angle mapping in human brain at 7T by multiple preparation
flip angles. Magn Reson Imaging 2020;72:71-77.
8. Rooney WD, Johnson G, Li X, Cohen ER,
Kim SG, Ugurbil K, Springer CS, Jr. Magnetic field and tissue dependencies of
human brain longitudinal 1H2O relaxation in vivo. Magn Reson Med
2007;57(2):308-318.