Arun Joseph1,2,3, Tobias Kober4,5,6, and Tom Hilbert4,5,6
1Advanced Clinical Imaging Technology, Siemens Healthcare AG, Bern, Switzerland, 2Translational Imaging Center, Sitem-Insel, Bern, Switzerland, 3Departments of Radiology and Biomedical Research, University of Bern, Bern, Switzerland, 4Advanced Clinical Imaging Technology, Siemens Healthcare AG, Lausanne, Switzerland, 5Department of Radiology, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland, 6LTS5, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
Synopsis
T2*-weighted imaging is an important diagnostic tool
to evaluate normal and pathological tissues due to its sensitivity to iron deposition.
This comes with high sensitivity to magnetic field inhomogeneities within a
voxel, resulting in susceptibility artifacts. These artefacts are largely reduced
at higher resolutions; high-resolution protocols however lead to clinically
unfeasible scan times with multi-echo gradient echo sequences. Here, we propose
a compressed sensing multi-echo GRE acquisition for 7T to obtain 0.8 mm isotropic
R2* maps of the whole brain in <7 minutes and with substantially reduced
artefacts. Preliminary qualitative and quantitative validations are performed
on healthy subjects.
Introduction
Quantitative R2* mapping can be an important predictor
of iron deposition in the brain and other organs of the human body1.
The accurate estimation of R2* maps are highly dependent on multi-echo T2*-weighted
acquisitions which are sensitive to local magnetic field inhomogeneities caused
by susceptibility differences (e.g. at air-tissue boundaries above the nasal
cavity) resulting in geometric distortion or signal voids2. One approach
to mitigate these artifacts is the reduction of intra-voxel dephasing through
acquisitions with smaller voxel size3. Imaging at high magnetic
field strength provides improved signal-to-noise ratios which can be used to achieve
higher resolutions and thus mitigate these artefacts. However, the acquisition
time increases drastically with higher resolution leading to impractical
experimental protocols.
Here, we propose to use a compressed sensing (CS)
multi-echo GRE sequence not to accelerate the acquisition but to trade the
gained acquisition time for higher resolution and thus significantly reduced
susceptibility artifacts. Methods
Four subjects were scanned at 7T (MAGNETOM Terra, Siemens Healthcare,
Erlangen, Germany) using a 1-channel transmit and 32-channel receive head coil (Nova
Medical, Wilmington, USA) after written informed consent was obtained. A
reference measurement with isotropic resolution of 1.6 mm was performed using a
product GRE sequence with GRAPPAx2 acceleration resulting in an acquisition
time of 6.02 min. A Cartesian spiral-phyllotaxis undersampling scheme
4
was implemented in a prototype GRE sequence. This prototype was then used to
acquire two additional datasets.
- An acquisition that has the
same resolution but a 5-fold CS acceleration (TA 1.42 min).
- An acquisition that uses a
5-fold acceleration but with a smaller voxel size (0.8 mm) to both approximately
match the acquisition time (TA 6.39 min) of the reference and reduce
susceptibility artifacts.
All measurements were performed in the sagittal plane with the following
scan parameters: TR 33 ms, TE 5/10/15/20/25/30 ms (6 echoes), FOV 256 mm
2,
flip angle 10°. An MP2RAGE sequence was acquired and used as anatomical
reference (TR 6000 ms, TE 2.87 ms, TI1/2 800/2700 ms, flip angles 4°/5°,1 mm
isotropic resolution).
Image reconstructions of all measurements were performed inline on the
scanner. While GRAPPAx2 measurements used the product reconstruction, CS
reconstructions were performed using a prototype iterative algorithm
5-7
with Haar wavelet regularization. The complex coil sensitivity maps for the CS
reconstruction were estimated using the ESPIRiT algorithm
8. The inline
reconstruction time for the CS reconstruction was ~5 minutes for one multi-echo
dataset. All post-processing and analysis of images were performed using Matlab
(MathWorks, USA). R2* maps were generated for each dataset by performing a
mono-exponential log-linear fit.
A region of interest analysis was performed on seven
white matter (WM) and gray matter (GM) structures (frontal-WM/GM, occipital-WM/GM,
parietal-WM/GM, and corpus collosum) using the prototype MorphoBox
9
segmentation algorithm on the MP2RAGE images. The resulting label maps were copied into the
space of the R2* maps using rigid registrations
10. The median values
from all regions were extracted using the obtained masks. Finally, the mean and
standard deviation across subjects for the different brain structures and
acquisitions were compared in a bar plot.
Results and Discussion
Figure 1 shows T2*-weighted images of the fifth echo obtained from the
GRAPPAx2 and CS acquisitions. The CS images show similar contrast to GRAPPAx2. Besides
the improvement in resolution, a drastic reduction of susceptibility artifacts can
be seen in the CS acquisition at higher resolution of 0.8 mm (see also red
arrows in Figure 1). Figure 2 shows the R2* maps generated from the multi-echo
GRE acquisitions. The R2* maps obtained from GRAPPAx2 and CS acquisitions were
found to be qualitatively comparable to each other. The R2* maps from CS acquisition
with 0.8 mm isotropic resolution had fewer susceptibility artifacts. Figure 3 shows
R2* maps from the transversal view for GRAPPAx2 and CS acquisitions. The effect
of high-resolution CS acquisition is demonstrated through clear delineation of mid
brain structures such as the substantia nigra. Figure 4 shows the distribution
of median values over three subjects for the different regions of interest and
acquisitions. The median values obtained from the R2* maps for GRAPPAx2 and CS
reconstruction were found to be similar. The median values obtained from CS
acquisition with 0.8 mm isotropic resolution was lower than the 1.6 mm
isotropic GRAPPAx2 and CS acquisitions. Figure 5 animates a comparison between
GRAPPAx2 and CS acquisitions with 1.6 mm and 0.8 mm resolution to highlight
difference due to resolution. Although both acquisitions have a similar scan
time of around 6 mins, a clear difference in image quality can be observed.Conclusion
We implemented CS acquisitions based on Cartesian
spiral-phyllotaxis readout scheme at 7T to generate 0.8 mm isotropic R2* maps
of the whole brain at a scan time of <7 mins. The preliminary data indicated that CS reconstruction
at higher resolution provided consistent data with drastically reduced susceptibility
artifacts and good qualitative information of the midbrain structures, which
enables its use in future clinical studies.Acknowledgements
No acknowledgement found.References
- Wood JC,
Enriquez C, Ghugre N, et al. MRI R2 and
R2* mapping accurately estimates hepatic iron concentration in
transfusion-dependent thalassemia and sickle cell disease patients. Blood 2005;106(4):1460-5.
- Chavhan
GB, Babyn PS, Thomas B, et al. Principles, Techniques, and Applications of
T2*-based MR Imaging and Its Special Applications. RadioGraphics. 2009;29 (5):
1433–1449.
- Petersa
AM, Brookesa MJ, Hoogenraadb FG, et al. T2* measurements in human brain at 1.5,
3 and 7 T. Magn Reson Imag. 2007; 25:748–753.
- Mussard E, Hilbert
T, Forman C, et al. Accelerated MP2RAGE
imaging using Cartesian phyllotaxis readout and compressed sensing
reconstruction. Magn Reason Med. 2020; 84(4):1881-1894.
- Forman C,
Piccini D, Grimm R, et al. High-resolution
3D whole-heart coronary MRA: a study on the combination of data acquisition in
multiple breath-holds and 1D residual respiratory motion compensation. MAGMA.
2014;27(5):435-43.
- Wetzl J, Forman C, Wintersperger BJ, et al. High-resolution dynamic CE-MRA of the thorax enabled
by iterative TWIST reconstruction. Magn Reson Med. 2017;77(2):833-840.
- Ma LE,
Markl M, Chow K, et al. Aortic 4D flow MRI in 2 minutes using compressed
sensing, respiratory controlled adaptive k-space reordering, and inline
reconstruction. Magn Reson Med. 2019;81(6):3675-3690.
- Uecker M,
Lai P, Murphy MJ, et al. ESPIRiT--an
eigenvalue approach to autocalibrating parallel MRI: where SENSE meets GRAPPA.
Magn Reson Med 2014;71(3):990-1001.
- Schmitter
D, Roche A, Maréchal B et al. An evaluation of volume-based morphometry for
prediction of mild cognitive impairment and Alzheimer's disease. Neuroimage
Clin. 2014; 7:7-17.
- Klein S, Staring M,
Murphy K, et al. "Elastix: a toolbox for intensity based
medical image registration," IEEE Transactions on Medical Imaging 2010;
29(1):196-205.