Huihui Ye1,2, Berkin Bilgic1, Stephen Cauley1, Borjan Gagoski3, Jianghui Zhong2, Yiping Du2, Lawrence L. Wald1, and Kawin Setsompop1
1MGH/A.A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, 2Zhejiang University, Hangzhou, China, People's Republic of, 3Boston Children's Hospital, Charlestown, MA, United States
Synopsis
Susceptibility weighted FLASH imaging provides
exquisite soft tissue contrast at high spatial resolution and low distortion,
but at a cost of lengthy acquisition time. In this work, we developed a highly
efficient Echo-Shift wave-CAIPI technique and demonstrated its ability to provide
18-fold acceleration for FLASH acquisitions with minimal noise and artifact
penalties. Instead of conventional single slice or slab echo-shift, we propose an
echo-shift Simultaneous MultiSlice method to enable improved controlled
aliasing and desirable volumetric noise averaging while avoiding slab edge
artifacts. With this technique, we demonstrated high-quality 1.5 mm isotropic
whole-brain susceptibility weighted FLASH at 3T and 7T in 36s. Purpose
The presence of
susceptibility shifts gives rise to a phase shift of the excited magnetization which
evolves linearly with time. Thus acquisitions at long TE(at ~T
2*)
are desirable for SWI and QSM. If TR is
kept >TE this leads to lengthy acquisition times. Echo-shift(ES) FLASH
1,2 , which efficiently interleaves 2D slice or slab
acquisitions and shifts the echo into the next TR period allows TR<TE and
improves efficiency. ES has also been used for other T
2* weighted
acquisitions such as fMRI(e.g. PRESTO)
3, and later combined with Simultaneous
MultiSlice(SMS)
4 for further acceleration.
We previously proposed improving
parallel imaging in FLASH using the Wave-CAIPI trajectory
5 which can provide an order of magnitude
acceleration in 3D FLASH acquisition with minimal g-factor and image artifact penalties. Wave-CAIPI uses an innovative controlled aliasing scheme
which takes full advantage of the available coil sensitivity information in
multi-channel receiver arrays. In this work, we combine the ES and
wave-CAIPI method, to provide high quality susceptibility weighted FLASH
acquisitions up to 18x faster than unaccelerated, non-echoshifted acquisitions.
To achieve better performance, we utilize echo-shift SMS with wave-CAIPI which enables
desirable volumetric noise averaging and optimal controlled aliasing
performance while avoiding slice profile boundary artifacts in ES multi-slab acquisition
2. Using ES wave-CAIPI, we demonstrate high
quality whole-brain susceptibility weighted gradient echo acquisition at 3T and
7T with 1.5mm isotropic voxel size in 36s.
Methods
Fig 1 shows the ES-SMS FLASH sequence with
echo shift gradients applied on both Gx and Gz gradients, wherein the ES factor
is 2 and MB-48 excitation RF pulses with 1.5mm slice thickness are used. In this sequence, the first RF excites the
odd slices of the 3D imaging volume with the signal readout echo-shifted to the
second TR, while the second RF excites the even slices with the signal readout
shifted to the third TR. The use of ES-SMS enables a comb of slices to be
acquired simultaneously, which enables desirable volumetric noise averaging
while avoids slice edge issues of ES multi-slab acquisitions. Moreover,
the combined use of ES-SMS with accelerated Wave-CAIPI acquisition enables controlled
aliasing to be performed across the whole imaging FOV rather than a partial FOV
in the slice-direction; thereby
providing optimal parallel imaging performance to achieve low noise
amplification and image artifact penalties.
To investigate the performance of the slice
profile in ES wave-CAIPI, Bloch simulation was used to calculate the
steady-state signal of the acquisition, where the effect of both the interleave
MB RF pulses and echo-shift gradients were accounted for (through summing up the
signal across multiple sub-voxel positions along x to obtain the slice profile
signal at a given z position).
A healthy subject was scanned with informed
consent at 3T and 7T with 32 channel head array to acquire ES wave-CAIPI FLASH
data. Imaging parameters were FOV 220x220x144mm and 1.5mm
3 iso
resolution, R
yxR
zxES=3x3x2, BW=90Hz/px, RF
duration=5ms, TBW=4, FA=10
o, TR
eff/TE
eff=47/35ms, Tacq=36s.
After the acquisition, two echo-shifted volumes were reconstructed separately using
standard wave-CAIPI reconstruction
5 and concatenated in an interleaved fashion to
generate full volumetric data. The resulting volumetric phase images were then processed
with 2D harmonic filtering to the remove the background phase component
6, and dipole inversion with Total Generalized
Variation (TGV) regularization
7 was applied to estimate the underlying susceptibility
distribution.
Results
Fig 2 shows the simulated slice profiles for
acquisition with and without echo shift demonstrating no cross-talk and minimal
slice profile change resulting from the echo-shifting. Fig 3 shows the high
quality magnitude, tissue phase and susceptibility images obtained from the ES
wave-CAIPI data at 1.5mm isotropic resolution in 36s, with high SNR, low
artifact and image distortion. At R
yxR
z=3x3
acceleration, wave-CAIPI has previously been demonstrated to result in very low
noise amplification with g-factor of close to one
2 and ES
acceleration does not incur a noise amplification penalty. Furthermore, with
adequately large echo shift gradients, no signal leakage and/or image strip
artifacts across slices were observed.
Discussions & Conclusions:
Echo shift method has been
effectively incorporated into the wave-CAIPI sequence to attain rapid, high
quality susceptibility mapping at 3T and 7T. This acquisition employed the ES-SMS
method to provide high SNR, good slice-profile and optimal controlled aliasing.
A relatively high MB factor of 48 is used, where VERSE’ing
8 was not needed due to the low FA used. Further
acquisition acceleration of up to and beyond 30x will be feasible with the use of
higher ES factors such as 3-4. This can be achieved through the use of higher
BW readout and/or longer TE acquisitions (which are more suitable for e.g. 3T
MRI with longer T
2*).
Acknowledgements
This work has been supported through the NIH NIBIB grants R01EB017219, R00EB012107, R01EB017337and P41EB015896.References
1. Moonen CTW,
Liu G, van Gelderen P, Sobering G. A fast gradient-recalled MRI technique with
increased sensitivity to dynamic susceptibility effects. Magn Reson Med.
1992;26:184-189.
2. Ma Y-J, Liu
W, Zhao X, et al. 3D interslab echo-shifted FLASH sequence for susceptibility
weighted imaging. Magn Reson Med. 2015;0(0):0.
3. Liu G,
Sobering G, Duyn JH, Moonen CTW. A functional MRI technique combining
principles of echo-shifting with a train of observations(PRESTO). Magn Reson
Med. 1993;30:764-768.
4. Boyacioglu R, Schulz J, Norris D. Multiband Echo Shifted EPI. In: Proc
Intl Soc Mag Reson Med_SMS Workshop. ; 2015.
5. Bilgic B,
Gagoski BA, Cauley SF, et al. Wave-CAIPI for highly accelerated 3D imaging. Magn
Reson Med. 2014;00(May):1-11. doi:10.1002/mrm.25347.
6. Kaaouana T,
de Rochefort L, Samaille T, et al. 2D harmonic filtering of MR phase images in
multicenter clinical setting: Toward a magnetic signature of cerebral
microbleeds. Neuroimage. 2014;104:287-300.
doi:10.1016/j.neuroimage.2014.08.024.
7. Bilgic B,
Chatnuntawech I, Langkammer C, Setsompop K. Sparse methods for Quantitative
Susceptibility Mapping. Proc SPIE. 2015;9597:959711.
doi:10.1117/12.2188535.
8. Conolly S,
Nishimura DG, Macovski A, Glover GH. Variable-rate selective excitation. J
Magn Reson. 1988;78(3):440-458.