Guido Buonincontri^{1}, Pedro A Gómez^{2}, Matteo Cencini^{1,3}, Mauro Costagli^{1}, Graziella Donatelli^{1}, Rolf F Schulte^{4}, and Michela Tosetti^{1}

When using ultra-high field MRI scanners (UHF, B0>= 7T), quantitative imaging is challenging due to B0 and B1+ non-uniformities. Magnetic resonance fingerprinting (MRF) represents a great opportunity for quantitative imaging at UHF as it can estimate these effects at the same time of the parameters of interest. Here, we compare two novel 3D SSFP MRF approaches, one based on a three-dimensional spiral projection acquisition and one using a radial acquisition in vivo at 7.0T. We estimate M0, T1, T2 and B1+ simultaneously at high resolution (1mm isotropic) within 6.5 minutes acquisition time.

Purpose:

Recent technological advances have improved on the effectiveness of ultra-high field MRI scanners (UHF, B0>=7T). Although new contrasts have been explored at high field strengths, common approaches are inherently non-quantitative, mostly due to the difficulties in dealing with B0 and B1+ non-uniformities, as well as a higher specific absorption rate (SAR). Magnetic resonance fingerprinting (MRF) represents a great opportunity for quantitative imaging at UHF as it does not require fields to be homogeneous. By including nuisance parameters into the signal encoding model it is possible to estimate these at the same time with the parameters of interest.

Here, we compare two novel 3D SSFP MRF approaches based on a three-dimensional radial and spiral projection k-space acquisitions, estimating M0, T1, T2 and B1+ simultaneously, demonstrating it in vivo at 7.0T.

*Acquisition parameters:*
We obtained images of a uniform
phantom (T1 600ms, T2 45ms) and a healthy volunteer using a Discovery MR950 7T MRI system (MR950, GE Healthcare, Milwaukee, WI,
USA) equipped with a 2-channel transmit / 32-channel receive head coil (Nova
Medical, Wilmington, MA, USA). The acquisition
sequence was based on SSFP MRF ^{1} (FOV 19.2x19.2x19.2cm3, 192x192x192
matrix), using hard pulses of 500 us length for excitation. We used a fixed TR of 8ms,
acquiring 56 repetitions of 880 frames preceded by a 10ms-long hyperbolic
secant adiabatic inversion pulse, using the flip angle and phase lists in Figure
1a. To increase the capability of the sequence to discriminate T2 and B1+
effects, we introduced radiofrequency spoiling in the first part of the
acquisition following Cloos et al^{2}.
For comparison of the B1+ values, in the phantom we acquired a standard 2D
Bloch-Siegert B1 map^{3} (TR=100ms,
TE=13ms, Flip Angle=30°, 4ms Fermi pulse, 2kHz off-resonance).

*3D
spiral projection trajectory:* Three dimensional spiral projections, as recently described by Castets et al^{4} and by Cao et al^{5} were
obtained rotating a variable density spiral with 704 readout points at 250kHz
sampling rate, 55 interleaves in-plane every 880 frames, and rotating the plane
by the golden angle for each frame, achieving a total of 49280 interleaves.
Each trajectory achieved zero-moment nulling in x, y and z after which a
spoiler in z was added achieving 4π dephasing across a 1-mm voxel (see figure 1b).

*3D Radial
trajectory:* For comparison with 3D spiral projections, we designed a matching 3D
radial trajectory consisting of 49280 spokes of 358 readout points at 250kHz
sampling rate, fully-sampling the 3D kspace for achieving 1mm isotropic
resolution. Each spoke was acquired symmetrically with respect to k-space
center and achieved zero-moment nulling in x, y and z, after which a spoiler in
z was added achieving 4π dephasing across a 1-mm voxel (see figure 1c). To
increase spatial and temporal incoherence, we randomly permutated the spoke
indexes using a uniform probability distribution.

For both k-space trajectories tested, prior to reconstruction and matching, k-space data were combined using SVD compression and adaptive coil combination was performed in the image domain.

All acquisitions tested at 7.0T achieved a specific absorption rate (SAR) lower than 2 W/Kg, showing that these are safe at ultra-high field. Acquisition time for both MRF acquisitions was 6.5 minutes.

Figure 2 shows the comparison of the MRF B1+ map from our MRF acquisitions with a standard 2D Bloch-Siegert map acquired in 7 minutes, showing good agreement.

Figure 3 shows representative parametric maps in one subject. It can be noted that T1 and T2 maps are free from B1+ artifacts. Three-dimensional spiral projections obtained clearer images, due to the more efficient sampling of the edges of k-space, when compared to 3D radial trajectories. However, despite the relatively short spiral readouts, the radial images achieved higher robustness to blurring artifacts and a higher anatomical accuracy.

[1] Jiang Y, Ma D, Seiberlich N, Gulani V, Griswold MA. MR fingerprinting using fast imaging with steady state precession (FISP) with spiral readout; Magn Reson Med. 2015 Dec;74(6):1621-31. doi: 10.1002/mrm.25559. Epub 2014 Dec 9.

[2] Martijn A. Cloos, Florian Knoll, Tiejun Zhao, Kai T. Block, Mary Bruno, Graham C. Wiggins, and Daniel K. Sodickson. Multiparametric imaging with heterogeneous radiofrequency fields. Nat Commun. 2016; 7: 12445.

[3] Sacolick LI, Wiesinger F, Hancu I, Vogel MW. B1 mapping by Bloch-Siegert shift. Magn Reson Med. 2010 May;63(5):1315-22. doi: 10.1002/mrm.22357.

[4] Charles R. Castets William Lefrançois Didier Wecker Emeline J. Ribot Aurélien J. Trotier Eric Thiaudière Jean‐Michel Franconi Sylvain Miraux. Fast 3D ultrashort echo‐time spiral projection imaging using golden‐angle: A flexible protocol for in vivo mouse imaging at high magnetic field; Magn Reson Med. 2017 May;77(5):1831-1840. doi: 10.1002/mrm.26263. Epub 2016 May 12.

[5] Xiaozhi Cao, Congyu Liao, Qing Li, Huihui Ye, Hongjian He, Jianhui Zhong. Fast 3D MR fingerprinting with spiral projection acquisition for whole brain quantification imaging; Proceedings of the 26th ISMRM annual meeting, Paris 2018. Program number 1017

Figure 1: A) Flip angle and phase lists used. The flip angle
list was preceded by an adiabatic inversion pulse. B) sequence diagram for our
3D spiral projection acquisition. C) Sequence diagram for our 3D radial
trajectories. A fixed spoiler was present in z, while spokes were acquired with
a randomized angle

Figure 2: Comparison between Bloch-Siegert B1+ mapping and
MRF B1+ mapping in a uniform phantom. There is good agreement between the
techniques. Both 3D MRF techniques produced
uniform T1 and T2 values within 10%.

Figure 3: Demonstration of 3D radial and 3D spiral MRF in
vivo. T1 and T2 values are free from B1+ inhomogeneities. Maps obtained from 3d
spiral projections were clearer from undersampling artifacts (as in circle),
but showed more blurring of tissue borders (as pointed by arrow).