David Feinberg1,2 and Liyong Chen1,2
1Advanced MRI Technologies, LLC, Berkeley, CA, United States, 2Helen Wills Neuroscience Institute, Univ of California, Berkeley, Berkeley, CA, United States
Synopsis
A new technique of simultaneous 4D flow
imaging acquired with simultaneous multiple 3D volumes (SMV) is presented. The
velocity measurements are compared to conventional 4D flow imaging and have
very good correspondence.
Introduction
4D Phase contrast (PC) MRI is currently a popular approach to
flow imaging1 in which quantitative velocity information is measured in a 3D
volume of the head or body. In conventional PC imaging the bipolar gradient
pulse is used for linear velocity encoding of signal phase2 and combined
with velocity phase maps3. Our previous work developed and evaluated simultaneous
multi-slice (SMS) in 2D PC flow imaging to
increase the number of 2D PC images without increasing the scan time4. Here we
develop simultaneous multiple 3D volume (SMV), multi-slab imaging, as a novel
4D PC flow imaging technique. Methods
A 3D FLASH GE sequence with cine acquisition and prospective cardiac
gating was modified to excite with multiband (MB) rf pulsesand bipolar velocity
gradient. Imaging was performed in 6 normal subjects with both
single-3D volume and SMV 2 slab volumes or 3 slabs in head and thorax. Our 3T
scanner (Siemens Trio) is located in a neuroscience center and had limited
coils available, a spine coil could only be used for thorax imaging which
reduced the spatial sensitivity and image quality in neck and chest imaging. A
32 channel coil was available for head imaging. Imaging parameters for cine 3D
SMV and conventional 3D GRE PC are:
TR=6.16ms, TE=3.92ms, FA Angle=15°, multiband rf pulse width=2.0ms,
IPAT acceleration factor=2. For brain, the matrix size and FOV=190x154mm2,
matrix=(128x64) slab thickness= 32mm, (16 x 2mm slice). For thorax
imaging: FOV=340x234mm2, matrix=(128x88), slab thickness= 48mm, (16 x 3mm
slice). Acquisitions evaluated 2 or 3 slabs (SMV=2 or 3) with a 100% slab
spacing, as shown in Figure 1. The interleaving of two SMV acquisitions gave
complete coverage with either 4 slabs or 6 slabs, respectively. Venc was either
80 cm/s for head or 160 cm/s for thorax.Results
Fig.
2 shows the spatial coverage in magnitude images in a single SMV-3 slab
acquisition of PC head and neck imaging. The total scan time was 8-9
minutes, depending on heart rate, for SMV-2 or SMV-3 and conventional SMV=1 4D
PC acquisitions. There was no loss in spatial resolution using SMV
acquisitions. Figs3 and 4 shows representative magnitude and phase maps in the
interleaved SMV-2 and SMV-3 acquisitions in the vessels with very good
match between the velocity curves from conventional single slab 4D PC and SMV
multi slab in head and thorax, Figs 5 and 7, respectively. Our scanner
only having a spine and neck array (combined 12 channels) but no anterior chest
coil array, the SNR and coil sensitivity limited the great vessel imaging.
There was expected slab profile relate signal loss and aliasing in the
outermost 2 slices of each slab. There is also signal reduction in the lower
slab in the neck as the 32 channel head coil had lower sensitivity in the
lower head and neck regions. In these experiments we did not overlap the
interleaved slabs to account for the outer slice losses, although this could
easily have been achieved, Figs. 3-4. There was high correspondence
in the measured velocity curves using multi-slab and conventional single slab
acquisitions, shown in Figs 5 and 7.Discussion
While not evaluated here, the use of thinner 3D
volumes with SMV will allow more consistent inflow of spins throughout the
width of the slab, whereas a thicker 3D slab covering the same region could
have reduced inflow depending on velocity and time of flight distance of blood
flow. There was no limitation due to SAR or peak rf power in these studies due
to the low flip angle and low MB factor of 2 or 3 used. The acceleration
factors and acquisition speed were not fully optimized in these initial
experiments, such as higher in-plane acceleration, compressed sensing and higher
SMV factor. The implementation of controlled aliasing will allow higher
accelerations and improved image quality. The implementation using proper coil
arrays may allow closer spaced slabs and higher SNR. One potential
disadvantage of the technique is the requirement of two acquisitions for
contiguous volume coverage.Conclusion
A new technique of 4D flow imaging using simultaneously
acquired multiple 3D volumes is presented and shows good correspondence in
quantitative velocity measurement to conventional 4D flow imaging. Acknowledgements
NIH R44MH112210References
1. Markl et al JMRI
2012. 2. Hahn J Geophys 1960 3. Feinberg
DA, Diagnostic Imaging,1984. 4. Feinberg DA, Chen L,page 320, ISMRM 2016.