Changyu Sun1, Yang Yang2, Xiaoying Cai 1, Sophia Cui1, Daniel Auger1, Michael Salerno1,2,3, and Frederick H. Epstein1,3
1Biomedical Engineering, University of Virginia, Charlottesville, VA, United States, 2Medicine, University of Virginia, Charlottesville, VA, United States, 3Radiology, University of Virginia, Charlottesville, VA, United States
Synopsis
Spiral
cine DENSE is an established method for imaging myocardial strain, however a
relatively long acquisition time is a limitation. We developed a simultaneous multislice (SMS)
method to accelerate spiral cine DENSE imaging.
For SMS excitation we employed CAIPIRINHA phase modulation of the
multiple slices. For the SMS
reconstruction, we implemented a modified iterative conjugate gradient
sensitivity encoding (CG-SENSE) method.
Simulations and experiments in phantoms show that a 10-15% error occurs
when imaging two slices simultaneously.
Two-slice volunteer SMS images showed close agreement with
separately-acquired single-slice images. SMS with CG-SENSE may be an effective means to
accelerate spiral cine DENSE.
Purpose
Spiral cine
displacement encoding with stimulated echoes (DENSE) is an established method
for imaging myocardial strain [1]. The
advantages of spiral cine DENSE are rapid strain analysis and higher spatial
resolution compared with myocardial tagging; however, a longer acquisition time
is a limitation [2]. Simultaneous
multislice (SMS) imaging has made a major impact in MRI of the brain. However, to date SMS has not seen widespread
application in cardiac MRI and there has not been extensive effort devoted to
reconstruction of SMS spiral data [3].
We sought to develop an SMS spiral cine DENSE method in order to
accelerate DENSE imaging. Methods
For SMS
DENSE RF excitation we employed CAIPIRINHA phase modulation of the multiple
slices [3]. For the spiral SMS reconstruction (Fig. 1), we implemented a
modified iterative conjugate gradient sensitivity encoding (CG-SENSE) method
based on a method previously employed for radial CAIPIRINHA [4]. This
method requires sensitivity maps for each slice, and these were acquired using
separate breathold acquisitions for each slice.
We used phantoms to evaluate these methods, including simulations of SMS
spiral DENSE using separately acquired slices, as well as true SMS acquisitions
where slices were acquired simultaneously.
We also evaluated the SMS spiral cine DENSE method in a human volunteer,
where the SMS images were compared to separately-acquired DENSE images at
matched slice locations. For phantom
and human scans, RF coils had 30-34 channels.
Results
Figure 2 shows the results of simulations of SMS
spiral DENSE using two separately acquired slices (Fig. 2A,B). Fig. 2C shows the simulated SMS image
reconstructed using NUFFT [5]. Fig. 2D,E
show the two images of different slices recovered after application of the
spiral CG-SENSE algorithm, and Figure 2F,G shows the difference between the
recovered images and the original images.
The average percent errors in signal intensity in homogeneous regions
were 13.66% and 11.62% for the two slices, respectively. The evaluation of two simultaneously acquired
phantom slices is shown in Figure 3.
Specifically, Figure 3A,B show the NUFFT and phase-demodulated NUFFT of
the SMS raw data, depicting initial estimates of the two separate slices. Figure 3C,D shows the two images of
different slices recovered after application of the spiral CG-SENSE algorithm,
and Figure 3E,F shows the difference between the recovered images and the
separately acquired images shown in Figure 2A, B. The average percent errors in homogeneous
regions were 12.03% and 9.94% for the two slices, respectively.
Results from a human volunteer are shown
in Figure 4. Specifically, for a
reference, separately-acquired spiral cine DENSE short-axis images at basal and
mid-ventricular locations are shown in Figure 4A and Figure 4B, and
CG-SENSE-recovered SMS spiral DENSE images at the same location are shown in
Figure 4A,B. Qualitatively, there is
good agreement between the SMS spiral DENSE images and the separately-acquired
spiral cine DENSE images. Similarly,
myocardial strain maps in Figure 5 show a close correspondence between the SMS
and separately acquired slices.
Discussion
SMS with CG-SENSE
may be an effective means to accelerate spiral cine DENSE imaging. Phantom data suggest that errors may be on
the order of only 10-15% when acquiring two slices simultaneously. Volunteer imaging showed promising initial results. Ongoing work includes acquiring and
recovering more than two slices simultaneously, autocalibration methods for
sensitivity maps, modifying the reconstruction to include a constraint related
to temporal sparsity in the cardiac phases dimension, and evaluating SMS spiral
DENSE more thoroughly in phantoms and volunteers.Acknowledgements
Research support from Siemens Healthineers.References
1. Zhong et al. MRM. 2010;64(4):1089-1097. 2. Chen et
al. JCMR, 2016; 18(1):38-51. 3. Yang et al. ISMRM, 2017; #3242. 4. Yutzy et al. MRM, 2011; 65(6): 1630-1637. 5. Fessler et al. IEEE T Signal Proces, 2003; 51(2): 560-574.