Daniel Polak^{1,2}, Kawin Setsompop^{1,3,4}, Stephen F. Cauley^{1,3}, Borjan A. Gagoski^{3,5}, Himanshu Batt^{6}, Florian Maier^{2}, Lawrence L. Wald^{1,3,4}, and Berkin Bilgic^{1,3}

We introduce a highly accelerated
T1-weighted MP-RAGE acquisition that utilizes a novel reordering scheme and Wave-CAIPI
encoding to retain high image quality.
R=9-fold accelerated in vivo MP-RAGE scans
were performed in 71 sec, with maximum and average g-factor of g_{max}=1.27
and g_{avg}=1.06. Compared to the state-of-the-art 2D-CAIPIRINHA method,
this is a factor of 4.6/1.4 improvement in g_{max}/g_{avg}. In
addition, we demonstrate a 57 sec acquisition at 7T with R=12-fold acceleration.
This acquisition had a g-factor performance of g_{max}=1.15 and g_{avg}=1.04.
Wave encoding overcomes the g-factor noise
amplification penalty and allows for an order of magnitude acceleration of
MP-RAGE acquisitions.

In conventional MP-RAGE acceleration along
the partition encoding only shortens the echo train length of the gradient echo
readout without affecting the acquisition time^{5}. For this reason, Wave-CAIPI
MP-RAGE utilizes a novel scheme (Fig. 1) that merges R_{z} planes of
k_{x}-k_{z} k-space in an interleaved fashion. This ensures
that the k-space center of each plane is acquired close to the inversion time
(TI) preserving the T1w contrast.

In compliance with IRB requirements a healthy
volunteer was scanned on a 3T Siemens Skyra scanner using a 32-channel product
coil. We achieved 69 sec scan time at R=3x3 acceleration with TE/TR/TI =
3.8/2500/1100 ms, CAIPI shift 1 and 11 sinusoidal cycles per readout at 8.8 mTm^{-1} maximum gradient amplitude. In addition, a 1.8 s calibration scan (GRE)
was acquired to compute the coil sensitivity profile using ESPIRiT^{6}.
A conventional 2D-CAIPI^{7} scan without wave gradients served as a
benchmark for comparison. The reconstruction of all presented datasets and
corresponding g-factor maps were performed in MATLAB.
To assess the quality of Wave-CAIPI in
comparison to techniques routinely used in clinical settings, we acquired three
averages of Wave-CAIPI using the protocol described above (total scan time 3
min 23 sec) and a R=4x1 GRAPPA measurement of similar duration (3 min 14 sec
including 24 integrated ACS lines).
Even higher acceleration for a 1mm isotropic
full brain scan was achieved on a Siemens Magnetom 7T scanner. At R=4x3
acceleration the measurement time was reduced to 57 sec. Due to physiological constraints
9 sinusoidal cycles and 16.0 mTm^{-1 }maximum gradient amplitude were chosen.
We also performed a series of g-factor
simulations for various number of sinusoidal cycles and maximum gradient slew
rates to investigate the wave parameter space. Furthermore, the effect of T1 recovery
and small flip angle excitation in Wave-CAIPI MP-RAGE (echo spacing 8 ms, flip
angle 9°, 10 mTm^{-1} maximum gradient amplitude) was simulated for a homogenous object
of T1=1500ms.

Figure 2 shows Wave-CAIPI and 2D-CAIPI acquisitions at R=3x3 acceleration with corresponding g-factor maps. While 2D-CAIPI suffered from severe noise amplification particularly in the brain stem, Wave-CAIPI demonstrated much enhanced encoding performance and 1.5/4.6-fold improvement in avg./max. g-factor.

Furthermore, a time matched R=4x1 GRAPPA scan was compared to three averages of R=3x3 Wave-CAIPI MP-RAGE (Fig. 3). Both acquisitions provided comparable image quality, SNR and T1-weighted contrast. However, Wave-CAIPI is anticipated to provide increased robustness to patient motion, as motion-corrupted averages can be discarded.

More than an order of magnitude acceleration (R=12) was achieved at 7T with corresponding scan time of 57 sec. Despite the high loss of SNR due to the intrinsic sqrt(R) penalty, all three views of Fig. 4 show detailed contrast and image quality.

Figure 5 depicts the results of the wave parameter investigation. It revealed that the g-factor is mainly determined by the maximum gradient amplitude and mostly independent of the number of cycles. However, increased number of cycles was found to reduce artefacts arising from T1 recovery (Fig. 5e).

[1] J. P. Mugler, J. R. Brookeman. Three-dimensional magnetization-prepared rapid gradient-echo imaging (3D MP RAGE). Magn. Reson. Med. 1990;15(1):152–157.

[2] K. P. Pruessmann, M. Weiger, M. B. Scheidegger, P. Boesiger. SENSE: Sensitivity encoding for fast MRI. Magn. Reson. Med. 1999;42(5):952–962.

[3] M. A. Griswold, P. M. Jakob, R. M. Heidemann, M. Nittka, V. Jellus, J. Wang, B. Kiefer, A. Haase. Generalized autocalibrating partially parallel acquisitions (GRAPPA). 2002;47(6):1202–1210.

[4] B. Bilgic, B. A. Gagoski, S. F. Cauley, A. P. Fan, J. R. Polimeni, P. E. Grant, L. L. Wald, and K. Setsompop. Wave-CAIPI for highly accelerated 3D imaging. Magn. Reson. Med. 2015;73(6):2152–2162.

[5] K. Setsompop, D. A. Feinberg, J. R. Polimeni. Rapid brain MRI acquisition techniques at ultra-high fields. NMR Biomed. 2016;29:1198-1221.

[6] M. Uecker, P. Lai, M. J. Murphy, P. Virtue, M. Elad, J. M. Pauly, S. S. Vasanawala, M. Lustig, ESPIRiT--an eigenvalue approach to autocalibrating parallel MRI: where SENSE meets GRAPPA. Magn. Reson. Med. 2014;71(3):990–1001.

[7] F. A. Breuer, M. Blaimer, M. F. Mueller, N. Seiberlich, R. M. Heidemann, M. A. Griswold, P. M. Jakob. Controlled Aliasing in Volumetric Parallel Imaging (2D CAIPIRINHA). 2006;55(3):549–556.

Figure 1: (left) Standard k-space sampling for R=3x3 acceleration using
CAIPI shift 1. Per TR one k_{x}-k_{z} plane is acquired, the
blurring in partition encoding direction is reduced, however the overall scan
time is only affected by the acceleration in phase encoding direction (R_{y} = 3).
(Right) Three k_{x}-k_{z} planes are merged to form one P-A-R
cycle, which shortens the scan time by a factor of 9. The reordering enforces
the k-space of each plane to be acquired at inversion time TI, preserving the
MP-RAGE contrast.

Figure 2: R=3x3 accelerated acquisitions of Wave-CAIPI and 2D-CAIPI are
compared at 3T (scan time 71 sec). Zoom-in (yellow) shows significant noise
amplification for 2D-CAIPI. The panel on the right, demonstrates inverse
g-factor maps and reports average and maximum g-factor values for Wave-CAIPI
and 2D-CAIPI.

Figure 3: Comparison three averages of R=3x3
accelerated Wave-CAIPI (scan time 3 min 23 sec) vs. R=4x1 GRAPPA (scan time 3 min
14 sec) at 3T. All three views show comparable image quality and negligible
artifacts.

Figure 4: R=4x3 accelerated Wave-CAIPI
acquisition at 7T (scan time 57 sec). Inverse g-factor and average/maximum
g-factor values depicted in the blue panel on the right.

Figure 5: a,b,c,d) g-factor simulations for
various number of sinusoidal cycles 3, 7, 11, 15, gradient slew rates 10, 20,
30, 40, 60, 80, 100, 120, 140, 160, 180 mTm^{-1}ms^{-1} and
corresponding gradient amplitude. For a given amplitude the g-factor was found
to be fairly independent of the number of cycles. e) Deblurring artefact (red
arrow) arising from small flip angle excitation and T1 recovery in Wave-CAIPI
MP-RAGE depicted for 3, 7, 11, 15 cycles and 10 mTm^{-1 }maximum gradient amplitude.
Artefact is much reduced for 11 and 15 cycles.