Feiyu Chen1, Valentina Taviani2, Joseph Y Cheng3, Tao Zhang4, Brian A Hargreaves3, John M Pauly1, and Shreyas S Vasanawala3
1Electrical Engineering, Stanford University, Stanford, CA, United States, 2Global MR Applications and Workflow, GE Healthcare, Menlo Park, CA, United States, 3Radiology, Stanford University, Stanford, CA, United States, 4Global MR Applications and Workflow, GE Healthcare, Houston, TX, United States
Synopsis
Wave encoding was implemented in a variable-density single-shot
fast spin echo (VD-SSFSE) pulse sequence. Auto-calibrated estimation of the
wave-encoding point-spread function (PSF) and coil sensitivity maps was used. Images
were reconstructed with parallel imaging and compressed sensing reconstruction.
Compared to non-wave-encoded Cartesian imaging, wave-encoded VD-SSFSE achieves improved
image quality with reduced aliasing artifacts at higher acceleration factors and
with full k-space coverage, providing fast acquisitions and clinically relevant
echo times.
Purpose
Single-shot fast spin echo (SSFSE) can provide
excellent T2 contrast and good motion-robustness for abdominal and pelvic MRI. Variable
density (VD) sampling and variable refocusing flip (VRF) angles along the echo
train1 have been shown to reduce scan time, enable compressed
sensing reconstruction, and allow full k-space coverage with clinically
relevant echo times. However, the maximum acceleration factor of variable
density sampling is usually limited by the number of available coil elements in
the phase-encoding (PE) direction, and therefore, residual aliasing and noise
amplification can be observed in highly accelerated VD-SSFSE scans. In this
work, we combine wave encoding2 with VD-SSFSE to enable higher
acceleration factors. By applying sinusoidal modulation in the PE direction
during readout, the proposed wave-encoded VD-SSFSE can reduce noise
amplification and residual aliasing artifacts without decreasing the scanning
efficiency of conventional VD-SSFSE scans.Methods
A sinusoidal wave-encoding gradient (5.5 cycles, 12 mT/m amplitude),
played out during the readout of each kx encoding line, was added to
a VD-SSFSE pulse sequence (Fig. 1a). VD under-sampling patterns with 20 central
PE lines of auto-calibration signals (ACS), 50-70 ky views, and an
effective acceleration factor of about 5 were used (Fig. 1b). Over-sampling of
1.6-2.0 in the frequency-encoding (FE) direction was used to account for voxel
spreading effects due to wave encoding2. The VRF schedule was controlled by prescribing first, minimum, and last
flip angles as well as the flip angle corresponding to the center of k-space. A
90° minimum flip angle was used to minimize signal loss due to cardiac
pulsation in the left lobe of the liver.
The under-sampled wave-encoded k-space was used to estimate the
wave-PSF3. Auto-calibrated estimation of coil sensitivity maps4
(ESPIRiT5), and CS-SENSE image reconstruction6 with $$$\ell$$$1-wavelet regularization were performed in MATLAB and C (the BART toolbox7).
Simulated g-factor maps of wave-encoded and non-wave-encoded acquisitions with
uniform under-sampling were compared using estimated coil sensitivity maps of a
32-channel torso coil (NeoCoil, Pewaukee, WI) for uniform sampling patterns.
Phantom and volunteer scans were performed with
Institutional Review Board approval and informed consent at 3T (GE MR750,
Waukesha, WI) using a 32-channel receive-only torso coil (NeoCoil, Pewaukee, WI)
with FE along R/L and PE along A/P. Conventional Cartesian acquisitions were
performed for comparison using the same sampling pattern and reconstruction
framework.
Results
With an acceleration factor of 4 (Fig. 2a), wave
encoding achieved an average g-factor of 1.2 and a peak g-factor of 1.5, while the
corresponding Cartesian acquisition resulted in a g-factor of 1.4 and a peak
g-factor of 2.5. Wave encoding also reduced the average g-factor from 1.9 to
1.5 and reduced the peak g-factor from 3.6 to 2.1 when a 5x acceleration was
used (Fig. 2b). Phantom (Fig. 3) and in-vivo images (Fig. 4 and 5) of wave-encoded
acquisitions showed reduced residual aliasing artifacts, less blurring, and
more structural details than conventional Cartesian acquisitions. Similar T2
contrast was observed in Cartesian and wave-encoded images. The reconstruction
time per slice was 10s for Cartesian and 20s for wave encoding, and the
calibration of wave-PSF took about 60s prior to reconstruction.Discussion
Wave encoding creates voxel spreading in the FE direction2.
Therefore, it makes use of coil sensitivities more efficiently and achieves
lower g-factors than Cartesian acquisitions. When incorporated in VD-SSFSE
readouts, wave encoding significantly improves the image quality at an
effective acceleration factors of about 5. The readout duration, echo-spacing,
and TE of wave-encoded acquisitions are within 3 ms of those of conventional
Cartesian acquisitions, which helps maintain T2 contrast. Auto-calibration of
the wave-PSF and the coil sensitivity maps using the under-sampled wave-encoded
k-space is intrinsically more robust to motion, which is critical for body
applications, and reduces the complexity of translating the proposed approach to
clinical practice.Conclusion
Compared to conventional non-wave-encoded Cartesian
imaging, the proposed wave-encoded sampling achieves improved image quality
with reduced aliasing artifacts at higher acceleration factors and with full
k-space coverage, thus providing fast acquisitions and clinically relevant echo
times. Combined with variable-density sampling and compressed sensing,
wave-encoded sampling enables acceleration factors of close to 5 for 2D SSFSE
scans.Acknowledgements
GE Healthcare, NIH R01 EB009690, NIH R01 EB019241, P41 EB015891.References
[1] Taviani V, Litwiller DV, Tamir JI,
Loening AM, Hargreaves BA, and Vasanawala SS. Variable Density Compressed
Sensing Single Shot Fast Spin Echo. Proc. Intl. Soc. Mag. Reson. Med. 2016; 24:
618.
[2] Bilgic B, Gagoski BA, Cauley SF, Fan AP,
Polimeni JR, Grant PE, Wald LL, Setsompop K. Wave-CAIPI for highly accelerated
3D imaging. Magnetic Resonance in Medicine. 2015; 73: 2152-2162.
[3] Chen F, Zhang T, Cheng JY, Pauly JM, and
Vasanawala SS. Auto-Calibrating Wave-CS for Motion-Robust Accelerated MRI.
Proc. Intl. Soc. Mag. Reson. Med. 2016; 24: 1857.
[4] Uecker M, Lai P, Murphy MJ, Virtue P,
Elad M, Pauly J, Vasanawala SS, Lustig M. ESPIRiT - An Eigenvalue Approach to
Autocalibrating Parallel MRI: Where SENSE meets GRAPPA. Magn Reson Med 2014;
71:990-1001.
[5] Curtis A, Bilgic B, Setsompop K, Menon RS, Anand CK. Wave-CS: Combining wave
encoding and compressed sensing. Proc. Intl. Soc. Mag. Reson. Med. 2015; 23:
0082.
[6] Cauley SF, Setsompop K, Bilgic B, Bhat H,
Gagoski B, Wald LL. Autocalibrated wave-CAIPI reconstruction; Joint optimization of k-space trajectory and parallel imaging reconstruction. Magnetic
Resonance in Medicine. 2016; Epub ahead of print.
[7] Uecker M, Ong F, Tamir JI, Bahri D, Virtue
P, Cheng JY, Zhang T, Lustig M. Berkeley Advanced Reconstruction Toolbox. Proc.
Intl. Soc. Mag. Reson. Med. 2015; 23:2486.