Suvai Gunasekaran1, Hassan Haji-Valizadeh2, Bradley Allen1, Ryan Avery1, and Daniel Kim1
1Northwestern University, Chicago, IL, United States, 2Harvard University, Boston, MA, United States
Synopsis
Current methods
for contrast-enhanced thoracic MR angiography (CE-MRA) utilize suboptimal imaging practices such as breath holding or long scan time.
Here we developed a 9.6-fold accelerated CE-MRA sequence using stack-of-stars
k-space sampling and XD-GRASP reconstruction to produce predictable scan time
(< 5 min), without significant loss in image quality. Our sequence was
significantly faster than the current clinical free-breathing scan, reducing
the scan time by 50%. Additionally, our
accelerated CE-MRA scan produced comparable image quality that was clinically
acceptable.
Introduction
Contrast-enhanced thoracic MR angiography (CE-MRA)
is routinely used to diagnose and evaluate aortic diseases including aneurysm,
dissection, and aortic valve related
aortopathy (1). Two common approaches to thoracic CE-MRA are breath-hold imaging with ECG
triggering and inversion-recovery navigator-gated (IR-NAV) CE-MRA. Disadvantages
of breath-held ECG-gated CE-MRA include low spatial resolution and spatial
coverage, susceptibility to arrhythmia related artifacts, and challenges with
breath-holding in some patients. IR-NAV CE-MRA can achieve higher spatial
resolution and coverage, but at the expense of scan time (up to 10 min) which
is also dependent on patient’s breathing pattern. The purpose of this study was
to develop and clinically evaluate a 9.6-fold accelerated IR CE-MRA sequence to
produce predictable scan time (< 5 min). Methods
Human Subjects:
Twenty patients (15 males, age
= 58 ± 17 years) who were undergoing a thoracic MRA for clinically indicated
reasons (7 bicuspid aortic valve, 6 aortic aneurysm, 5 aortic dilation, 2
aortic valve replacement) were consented to participate in our study. Patients
underwent clinical breath-held ECG-gated
CE-MRA (0.15‐0.2
mmol/kg of gadobutrol) and randomized to undergo
either the clinical IR-NAV CE-MRA (n=10) or our accelerated IR CE-MRA (n=10).
Pulse Sequence: We modified a previously described non-contrast thoracic
MRA pulse acquisition with XD-GRASP reconstruction (2) . First, for each shot of 48 rays along the partition
direction, rays were rotated with tiny golden angle (TGA) = 23.6281° (3) (Figure 1). Second,
the k-space trajectory was designed with a fixed inversion time (TI) to achieve
consistent T1 weighting. Relevant
imaging parameters for the three CE-MRA sequences used are summarized in
Table 1.
Image Reconstruction: XD-GRASP data were reconstructed as previously described (2). Briefly, data were
rebinned to 6 respiratory states using principle component analysis of the navigator echo signal. Temporal total variation was used as the sparsifying
transform with normalized regularization weight = 0.00075 and fidelity = 0.002 (22
iterations, conjugate gradient with backtracking line search). During post-processing,
we applied low rank block wise (4) filtering with 2 iterations using a block size
of 8 with a weight of 0.15 and a fidelity weight of 0.01.
Data Analysis: Two cardiovascular radiologists with 8 years and 15 years of clinical
experience with MRA graded the image quality. In total, 40 CE-MRA image sets
(20 for the breath-held
ECG-gated CE-MRA, 10 for the IR-NAV CE-MRA, and 10 for the accelerated IR
CE-MRA) from 20 patients were randomized and de-identified for display on a
DICOM viewer (RadiAnt DICOM Viewer, Medixant, Poznan, Poland). Prior to visual
evaluation, the readers were given training data sets with poor to excellent
quality to calibrate reader’s scores in consensus. Following this training
session, the readers independently graded the scores by being blinded to
image type, each other’s score, and clinical history. The readers graded
for each of three categories on a 5-point Likert scale: conspicuity of
vasculature (1: nondiagnostic; 2: poor; 3: clinically acceptable; 4: good; 5:
excellent), noise and artifact levels (1: nondiagnostic; 2: severe; 3:
moderate; 4: mild; 5: minimal). The summed visual scores (SVS) was calculated
as the sum of three scores, where 9 was defined as clinically acceptable. Results
The
clinical breath held scan took on average 34 ± 7 seconds and was run immediately after contrast agent administration. The clinical IR-NAV CE-MRA
scan was run on average at 2.1 ± 0.8 min after contrast agent administration
and took 6.4 ± 2.6 min to complete the scan. Our accelerated IR CE-MRA was run on
average 1.3 ± 0.7 min after contrast agent administration and took 3.3 ± 0.5
min to complete the scan. Figure 2 shows representative examples from two
patients, illustrating comparable image quality relative to breath-hold CE-MRA
as an internal control. Figure 3 shows additional examples from six different
patients illustrating image quality between the clinical IR-NAV CE-MRA and our
accelerated IR CE-MRA. Upon visual inspection of all 20 patients, the median
conspicuity and artifact scores were not significantly different, whereas there
were significant differences for noise and SVS (Table 2). Nevertheless, the SVS
was greater than 9.0 for all CE-MRA methods. Discussion
Our
9.6-fold accelerated IR CE-MRA was significantly faster than the clinical
IR-NAV CE-MRA (p = 0.002) with average scan times of 3.3 ± 0.5 min and 6.4 ± 2.6 min respectively,
which represents a ~50% reduction in scan time. Our accelerated IR CE-MRA scan
produced comparable image quality that was clinically acceptable. While the
accelerated imaging was adequate for the clinical task of visualizing the
vasculature, the imaging suffers from higher noise levels compared to the
clinical breath-held and clinical IR-NAV scans. One weakness of this study is
that the imaging was unpaired between the IR-NAV CE-MRA and the accelerated IR
CE-MRA. While it would have been ideal to run both the IR-NAV CE-MRA and the
accelerated IR CE-MRA in the same patient, running them back to back inherently
reduces the image quality for the second scan run due to gadolinium washout in
the blood pool. Therefore, we elected to use the breath-held ECG-gated CE-MRA
as an internal control for both sequences. Conclusion
Accelerated IR CE-MRA significantly reduced scan time by half compared to
the clinical IR-NAV CE-MRA while maintaining clinically acceptable image
quality. Acknowledgements
Funding: National Institutes of Health
(R01HL116895, R01HL138578, R21EB024315, R21AG055954, T32EB025766) and American
Heart Association (19IPLOI34760317). References
1. Ho VB, Corse WR, Hood MN, Rowedder
AM. MRA of the thoracic vessels. Semin Ultrasound CT MR. 2003;24(4):192-216.
2. Haji-Valizadeh H, Collins JD, Aouad
PJ, Serhal AM, Lindley MD, Pang J, et al. Accelerated, free-breathing,
noncontrast, electrocardiograph-triggered, thoracic MR angiography with
stack-of-stars k-space sampling and GRASP reconstruction. Magn Reson Med.
2019;81(1):524-32.
3. Wundrak S, Paul J, Ulrici J, Hell E,
Geibel MA, Bernhardt P, et al. Golden ratio sparse MRI using tiny golden
angles. Magn Reson Med. 2016;75(6):2372-8.
4. Ong
F, Lustig M. Beyond Low Rank + Sparse: Multi-scale Low Rank Matrix Decomposition.
IEEE J Sel Top Signal Process. 2016;10(4):672-87.