Zheng Zhong1,2, Qingfei Luo1, Kaibao Sun1, Guangyu Dan1,2, Muge Karaman1,2, and Xiaohong Joe Zhou1,2,3,4
1CMRR, University of Illinois at Chicago, Chicago, IL, United States, 2Bioengineering, University of Illinois at Chicago, Chicago, IL, United States, 3Neurosurgery, University of Illinois at Chicago, Chicago, IL, United States, 4Radiology, University of Illinois at Chicago, Chicago, IL, United States
Synopsis
An imaging technique, coined epi-based Sub-millisecond Periodic Event
Encoded Dynamic Imaging, or epi-SPEEDI, has been shown capable of visualizing
the rapid opening and closing of human aortic valve with sub-millisecond
temporal resolution. However, the acquisition time of epi-SPEEDI was relatively
long (8-10 breath holds). Herein, we report an alternative SPEEDI technique
with a reduced FOV, which we call rFOV-SPEEDI, to shorten the scan times. This
technique has been successfully applied to visualization of human aortic valve
opening and closing with a temporal resolution of 0.6 ms, together with a scan
time reduction of 32.5%.
Introduction
Visualization
of the rapid opening and closing of the human aortic valve is important in
detecting valvular diseases such as stenosis and regurgitation. This has been
clinically done using ultrasound echography due to the limited temporal resolution of MRI and
the rapid movement of valvular structures. Recently, a novel MRI technique, coined epi-based
Sub-millisecond Periodic Event Encoded Dynamic Imaging, or epi-SPEEDI1, was proposed and successfully
used to capture the dynamic process of aortic valve opening and closing with a
temporal resolution of 0.6ms. However, the acquisition time of epi-SPEEDI is
relatively long (8-10 breath holds), which can potentially introduce
position inconsistency among different breath holds. One approach to shortening
the acquisition time is to limit the field-of-view (rFOV), thereby reducing the
number of phase-encoding steps while maintaining the spatial resolution, as
demonstrated in other applications2–4. This approach is
particularly suitable for aortic valve imaging where only a small FOV is needed.
Herein, we report an rFOV SPEEDI technique, which we call rFOV-SPEEDI, to
shorten the scan times and demonstrate its ability to visualize the human aortic valve
opening and closing with a sub-millisecond temporal resolution. Methods
rFOV-SPEEDI
Sequence
rFOV-SPEEDI
is built upon an epi-SPEEDI pulse sequence reported recently1. By synchronizing with a
cyclic event under investigation (Figure 1), N gradient echoes from the
echo-train in epi-SPEEDI are employed to sample N separate k-space matrices,
each corresponding to a distinct time point. The echo-train acquisition is
repeated M times, producing N resultant k-space matrices that
are individually reconstructed to generate a series of time-resolved images.
The temporal resolution is determined by the echo spacing (esp), which is
typically shorter than 1ms. By replacing the conventional excitation RF pulse
in epi-SPEEDI with a custom 2D RF pulse, rFOV-SPEEDI was able to reduce
the scan times by limiting the FOV and decreasing the number of required
phase-encoding steps (or M) accordingly.
Data Acquisition and Image Analysis
With IRB approval,
aortic valve images were acquired from healthy subjects on a 3T GE MR750
scanner using both rFOV-SPEEDI and full FOV epi-SPEEDI with EKG signals as a trigger. To capture the entire opening and closing processes of the aortic
valve, an “interleaved multi-phase” acquisition scheme was used with multiple trigger
delays as shown in
Figure 2. The temporal
gaps in-between the acquisition blocks were filled up using L
acquisition blocks acquired from different trigger delays (e.g., the blue
acquisition blocks can be dovetailed by the red acquisition blocks). Three trigger
delays (18ms/28ms/38ms) were used in the rFOV-SPEEDI scan with a total acquisition
time of 108 heartbeats in 5-6 breath-holds. The other key sequence parameters
for rFOV-SPEEDI acquisition were: TR/TE=29/8.8ms, flip angle = 10º, slice
thickness=8mm, FOV=12cm×12cm, matrix size=60×60, esp=0.6ms, L=20 per R-R
interval, N=16 per excitation, and M=36. For full FOV
acquisition, the imaging parameters were identical to those used in the rFOV-SPEEDI
sequence, except for TR/TE=20/3ms, FOV=24cm×24cm, matrix size=118×118, L=32,
M=80, trigger delays=12ms/22ms, and acquisition time = 160 heartbeats
in 8-10 breath-holds.
The acquired k-space data from both acquisitions was reconstructed offline using
customized Matlab programs. The reconstructed images were reordered according to their
acquisition time based on the acquisition strategies described in Figure 2. To
monitor the dynamic change of the aortic valve, the planimetric
aortic valve area
(AVA) was extracted from each image in the time series. Results
Figures 3 and 4 show two sets of time-resolved images, both with a temporal
resolution of 0.6 ms, during the rapid opening and rapid closing phases of the aortic
valve from a representative healthy human subject (male; 30-years-old). The dynamics were clearly visualized
from both full FOV and reduced FOV acquisitions. Compared with the full FOV
acquisition, rFOV-SPEEDI reduced the scan time by 32.5% (108 vs. 160 heartbeats) without compromising the image quality.
Figure 5 displays two planimetric AVA dynamic curves
obtained from the full FOV acquisition and reduced FOV acquisition, both with the temporal resolution of 0.6ms, together with the ECG waveform from the subject.
The two AVA curves matched well with each other. From either AVA curve, the
three phases during the aortic valve opening and closing can be well
identified: a rapid opening phase, a slow closing phase, and a rapid closing
phase. In addition, both AVA curves revealed an overshoot when the aortic valve
opened maximally, which was reported in the literature using non-MRI techniques5,6.Discussion and Conclusion
Using rFOV-SPEEDI,
we have experimentally observed the dynamics of aortic valve rapid opening and
closing in human subjects with sub-millisecond temporal resolutions and
adequate spatial resolutions in a reduced scan time of 108 heartbeats. This scan
time can be shortened further by combining rFOV-SPEEDI with parallel imaging
such as SENSE7, GRAPPA8, low-rank9, sparse matrix
decomposition10, machine learning11, and other sparse k-space
sampling methods. Based on the demonstration herein, the rFOV-SPEEDI technique may
find other applications for visualizing ultrafast and cyclic biological and
physical processes, especially in situations where the region of interest is in
a focal area within a larger object.Acknowledgements
This work was supported in part by a grant
from the National Institutes of Health (Grant No. NIH 1S10RR028898). The
content is solely the responsibility of the authors and does not necessarily
represent the official views of the National Institutes of Health. The authors
are grateful to Dr. Yi Sui of Mayo Clinic for initially designing the 2D RF
pulse.References
1. Zhong,
Z., Sun, K., Dan, G., Karaman, M. M. and Zhou, X. J. Capture the Opening and
Closing of Human Aortic Valve Using MRI with Sub-Millisecond Temporal
Resolution. in Proc. Intl. Soc. Magn. Reson. Med. (2020).
2. Zhong, Z. et al. High-Spatial-Resolution Diffusion
MRI in Parkinson Disease: Lateral Asymmetry of the Substantia Nigra. Radiology
291, 149–157 (2019).
3. Finsterbusch, J. Fast-spin-echo imaging of inner
fields-of-view with 2D-selective RF excitations. J. Magn. Reson. Imaging
31, 1530–1537 (2010).
4. Zhou, X. J. and Sui, Y. Image Domain Segmented Echo Planar
Magnetic Resonance Imaging Using a 2D Excitation Radiofrequency Pulse. US
Patent No. 9797970 (2016).
5. Laniado, S., Yellin, E., Terdiman, R., Meytes, I. and Stadler, J. Hemodynamic correlates of the normal aortic valve echogram. A study
of sound, flow, and motion. Circulation 54, 729–737 (1976).
6. van Steenhoven, A. A., Veenstra, P. C. and Reneman, R. S.
The effect of some hemodynamic factors on the behaviour of the aortic valve. J.
Biomech. 15, 941–950 (1982).
7. Pruessmann, K. P., Weiger, M., Scheidegger, M. B. and Boesiger, P. SENSE: sensitivity encoding for fast MRI. Magn. Reson. Med.
42, 952–962 (1999).
8. Griswold, M. A. et al. Generalized autocalibrating
partially parallel acquisitions (GRAPPA). Magn. Reson. Med. 47,
1202–1210 (2002).
9. Lingala, S. G., Hu, Y., DiBella, E. and Jacob, M.
Accelerated Dynamic MRI Exploiting Sparsity and Low-Rank Structure: k-t SLR. IEEE
Trans. Med. Imaging 30, 1042–1054 (2011).
10. Otazo, R., Candès, E. and Sodickson, D.
K. Low-rank plus sparse matrix decomposition for accelerated dynamic MRI with
separation of background and dynamic components: L+S Reconstruction. Magn.
Reson. Med. 73, 1125–1136 (2015).
11. Han, Y., Sunwoo, L. and Ye, J. C.
k-Space Deep Learning for Accelerated MRI. ArXiv180503779 Cs Stat
(2019).