Professional Documents
Culture Documents
Blaimer Et Al. - 2006 - Accelerated Volumetric MRI With A SENSEGRAPPA Com
Blaimer Et Al. - 2006 - Accelerated Volumetric MRI With A SENSEGRAPPA Com
Technical Note
difficult to extract the coil sensitivity information, such The aliased partitions can be unfolded by using fully
as in lung or abdomen imaging. Fourier-encoded low-resolution reference images to cal-
culate the reconstruction parameters. The reference
images can be acquired in an additional experiment
MATERIALS AND METHODS
before or after the accelerated scan or during the scan
Image Reconstructions by using a variable density (VD) acquisition scheme (9).
In 3D MRI phase encoding is performed along two spa- For each coil the corresponding low-resolution refer-
tial directions: the PE direction and the PAE direction. ence images 1 and 2 at different positions z1 and z2 are
Therefore, undersampling can be carried out along both combined to form an extended 2D image matrix ⬘. The
dimensions. A reduction in the sampling density under low-resolution reference image of the first partition is
the Nyquist criterion along the PE direction results in placed in the upper half of the extended matrix,
well known aliasing artifacts, which can be removed by
using pMRI methods. reordering
After the GRAPPA reconstruction and inverse Fourier langen, Germany). Healthy volunteers were examined,
transformation, single-coil images containing the sep- and informed consent was obtained before each study.
arated partitions are available (shown on the right side To investigate the performance of the SENSE/
of Fig. 2b). To obtain the final images, the single-coil GRAPPA combination, a fully-sampled 3D data set was
images can be combined, for example by using a sum- acquired using an eight-channel head coil array (MRI
of-squares reconstruction (14). After the reconstruction Devices, Waukesha, WI, USA) with a T1-weighted mag-
process, the individual partitions are reordered to their netization prepared rapid gradient-echo sequence (TR/
corresponding position. TE/FA ⫽ 1630 msec/4.4 msec/15°, FOV ⫽ 220 ⫻ 220
The reconstruction procedure described above must mm2, slab thickness ⫽ 144 mm, 72 partitions, matrix ⫽
be applied to all aliased partitions in the undersampled 256 ⫻ 256). Image reconstruction was performed offline
data set. Although GRAPPA is applied for image recon- using Matlab (The Mathworks, Natick, MA, USA). The
struction, effectively the reconstruction algorithm is fully-sampled data set was reduced by a factor of RPE in
SENSE along the partition direction. In general, the the PE direction (anterior–posterior) and by a factor of
SENSE reconstruction can be characterized as an im- RPAE in the partition direction (left–right), resulting in a
age domain unfolding algorithm. Aliased pixels are un- total reduction factor of R ⫽ RPE ⫻ RPAE. Low-resolution
folded using the knowledge of the complex coil sensitiv- reference images were obtained from the fully-sampled
ities at the corresponding positions. In the case of the data by extracting a matrix block, typically 48 parti-
SENSE/GRAPPA combination, the partition direction is tions with matrix size ⫽ 32 ⫻ 256, from the central
in the image domain. The aliased partitions are un-
k-space. The reference data were only used to calculate
folded partition-wise using a GRAPPA reconstruction
the reconstruction parameters and were not included
algorithm and then reordered to their corresponding
in the final image reconstruction.
position. Therefore, the reconstruction can be seen as
Additionally, a 2D-SENSE reconstruction (13) was
GRAPPA in the PE direction and SENSE in the partition
performed for comparison. The coil sensitivity maps
direction.
were calculated by using coil array correlation statistics
(15). To ensure accurate coil sensitivity maps, the size
In Vivo Experiments of the reference data was chosen to be 64 ⫻ 256, 64
Several in vivo imaging experiments were performed on partitions.
a 3.0 T whole-body MR scanner equipped with 32-chan- An additional accelerated experiment was performed
nel receiver technology (Siemens Medical Solutions, Er- using a prototype 32-channel receiver head coil array
3D MRI With a SENSE/GRAPPA Combination 447
Figure 3. Results from the 3.0 T whole-body system equipped with an eight-element head-coil array. Two representative
partitions from the fully-sampled data set are shown (left side). The same partitions reconstructed with the SENSE/GRAPPA
combination with acceleration factor R ⫽ RPAE ⫻ RPE ⫽ 2 ⫻ 2 ⫽ 4 (middle) and acceleration factor R ⫽ RPAE ⫻ RPE ⫽ 2 ⫻ 3 ⫽ 6
(right side) are shown.
(16) and a T1-weighted gradient-echo sequence (TR/ shown. Images with total acceleration factors of R ⫽ 4
TE/FA ⫽ 12 msec/4.7 msec/15°, FOV ⫽ 256 ⫻ 256 (Fig. 3, middle) and R ⫽ 6 (Fig. 3, right side) are pre-
mm2, slab thickness ⫽ 160 mm, 80 partitions, matrix ⫽ sented. On the visual scale the images are of good qual-
256 ⫻ 256). The acquisition was accelerated in the PE ity. No artifacts were introduced by the reconstruction.
direction by a factor of RPE ⫽ 3.2 and in the PAE direc- For comparison, the corresponding fully Fourier-en-
tion by a factor of RPAE ⫽ 3, resulting in a total acceler- coded images are shown (Fig. 3, left side).
ation factor of R ⫽ 9.6. For calibration of the recon- A comparison of the SENSE/GRAPPA combination
struction process, a low-resolution reference data set is with 2D-SENSE at six-fold acceleration is shown in Fig.
necessary. In this case, a fast spin-density-weighted 3D 4. The 2D-SENSE approach (Fig. 4a) is based on SENSE
fast low-angle shot (FLASH) experiment (matrix ⫽ 48 ⫻ and therefore results in the best possible reconstruc-
64, 36 partitions, acquisition time ⫽ 6 seconds) was tion with optimized SNR (10). The SENSE/GRAPPA
performed prior to the accelerated experiment to ac- combination (Fig. 4b) is based on the GRAPPA approach
quire the reference data (17). The total experiment time and is a close approximation of the optimum recon-
for this examination, including the acquisition of the struction. In this example, on the visual scale, the im-
reference data, was 31 seconds. The reader should note ages obtained with the SENSE/GRAPPA combination
that different contrasts were successfully used in the are of similar quality compared to the results from the
actual accelerated experiment and the low-resolution 2D-SENSE approach. Figure 4c provides a subtraction
reference data. of the 2D-SENSE result from the SENSE/GRAPPA re-
For all experiments a modified GRAPPA algorithm was sult for the same partition in order to make differences
used for image reconstruction (18,19). In our implemen- visible. For better visualization the subtraction image
tation a kernel of four neighboring points along the PE was multiplied by a factor of 5. The g-factor map for this
direction and three neighboring points along the frequen- partition (Fig. 4d) indicates that a six-fold acceleration
cy-encoding direction from all coils were combined to re- is marginal for this eight-channel coil array since the
construct a missing data point in a single coil. SNR is decreased by a factor of 7 in some regions within
the head.
In Fig. 5 images from a T1-weighted gradient-echo
RESULTS
experiment are shown in sagittal, transverse, and coro-
In Fig. 3 two representative partitions after pMRI recon- nal views. Using the prototype 32-channel head-coil
struction using the SENSE/GRAPPA combination are array, an acceleration factor of R ⫽ 9.6 was achieved.
448 Blaimer et al.
This example demonstrates that it is feasible to use the proach presented here splits the 2D reconstruction
SENSE/GRAPPA combination with high acceleration process into multiple 1D reconstruction processes.
factors. To every partition of the undersampled data set a
standard 1D-GRAPPA algorithm is applied in order to
unfold the aliased partitions. Finally, the individual
DISCUSSION
partitions are reordered at their corresponding posi-
In this work a combination of the SENSE and tions. Effectively, the reconstruction is GRAPPA along
GRAPPA algorithms is presented. Similarly to 2D- the PE direction and SENSE along the partition di-
SENSE, undersampling is performed in two dimen- rection, because aliased pixels are resolved partition
sions in order to improve the reconstruction quality of by partition. As in normal SENSE, the acceleration
3D imaging experiments. The computational ap- factor in PAE direction can be a non-integer number.
Figure 5. Experimental results obtained with a prototype 32-channel head-coil array. Transverse (left), sagittal (middle), and
coronal (right) views of a T1-weighted 3D gradient-echo experiment are shown. The experiment was accelerated in the PE
direction (anterior–posterior) by a factor of RPE ⫽ 3.2, and the PAE direction (left–right) by a factor of RPAE ⫽ 3, resulting in a total
acceleration factor of R ⫽ 9.6. Low-resolution spin-density-weighted reference images for calculating the reconstruction
parameters were acquired in an additional experiment prior to the actual experiment. The total experiment time was 31 seconds,
including the acquisition of the low-resolution reference images.
3D MRI With a SENSE/GRAPPA Combination 449
For example, a 1.5-fold acceleration in this direction the final reconstruction, which will improve the image
results in a superposition of two partitions with a quality by reducing the residual artifact power and
distance of 0.75 䡠 ⌬S from each other, with the slab increasing the SNR. A drawback of this strategy is the
thickness ⌬S. The superimposed partitions can be decrease in data acquisition efficiency because the
separated if corresponding reference images are total acquisition time is increased.
available. In this example, only a fraction of the data The SENSE/GRAPPA combination is easy to imple-
would have to be unfolded, since not all partitions ment and allows the use of existing reconstruction
suffer from aliasing. algorithms for 2D imaging. In principle, any other
The SENSE/GRAPPA combination is based on the current pMRI technique, such as SENSE or SMASH,
GRAPPA method and allows one to take advantage of its can be applied instead of the GRAPPA algorithm,
specific benefits. The reconstruction does not require which makes this computational approach very flex-
coil sensitivity maps and therefore the SENSE/ ible. Since a standard GRAPPA algorithm for 2D im-
GRAPPA-combination is advantageous for applications aging is used in our implementation, the computation
in which it is difficult to obtain sensitivity maps. Espe- time for a 3D imaging experiment can be roughly
cially in inhomogeneous regions with low spin density, estimated as the GRAPPA reconstruction time for a
such as the lung or abdomen, and in EPI experiments in 2D image multiplied by the number of partitions. The
regions with severe distortions, the GRAPPA method computation time for reordering the reference images
has been shown to be quite successful (see for example is negligible.
Ref. 20). Therefore, the SENSE/GRAPPA-combination In conclusion, the presented SENSE/GRAPPA com-
should be the method of choice whenever acceleration bination for 3D imaging allows one to take advantage
in two dimensions is desired for 3D extensions of these of the specific benefits of the normal GRAPPA method
applications.
in combination with 2D undersampling. The combi-
In this work we present the results from head im-
nation with a SENSE-like reconstruction along the
aging experiments. Using a 32-channel head-coil ar-
PAE direction reduces the complexity of the compu-
ray, the scan time for whole-brain coverage with an
tation.
in-plane resolution of 1 ⫻ 1 mm2 and 2 mm slice
thickness was reduced from about four minutes with-
out acceleration by pMRI down to 31 seconds with
parallel imaging, including the acquisition of the ref- REFERENCES
erence data.
As in all pMRI methods, reference measurements 1. Carlson JW. An algorithm for NMR imaging reconstruction based
for calibration of the reconstruction process are nec- on multiple RF receiver coils. J Magn Reson 1987;74:376 –380.
essary. In this work an additional low-resolution 3D 2. Hutchinson M, Raff U. Fast MRI data acquisition using multiple
detectors. Magn Reson Med 1988;6:87–91.
prescan was used to acquire the reference data. In
3. Kelton JR, Magin RL, Wright SM. An algorithm for rapid image
principle, a prescan with any arbitrary imaging con- acquisition using multiple receiver coils. In: Proceedings of the 8th
trast can be used for the GRAPPA reconstruction of Annual Meeting of SMRM, Amsterdam, The Netherlands, 1989.
pMRI experiments with totally different imaging con- Abstract 1172.
trasts (17). In this work we chose a fast spin-density- 4. Carlson JW, Minemura T. Imaging time reduction through multiple
receiver coil data acquisition and image reconstruction. Magn Re-
weighted 3D gradient-echo sequence for the acquisi-
son Med 1993;29:681– 687.
tion of the reference data. Thus, the acquisition time 5. Ra JB, Rim CY. Fast imaging using subencoding data sets from
for the prescan was minimized and did not become multiple detectors. Magn Reson Med 1993;30:142–145.
comparable to the scan time of the accelerated exper- 6. Kwiat D, Einav S. Preliminary experimental evaluation of an
iment. The resolution of the prescan data should be inverse source imaging procedure using a decoupled coil detector
large enough to appropriately register any coil sensi- array in magnetic resonance imaging. Med Eng Phys 1995;17:
257–263.
tivity variations along the PAE direction and to avoid 7. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial
Gibbs’ artifacts introduced by a Fourier transform of harmonics (SMASH): fast imaging with radiofrequency coil arrays.
the low-resolution reference data. In our experience, Magn Reson Med 1997;38:591– 603.
32– 48 partitions with a matrix size of 32 ⫻ 64 are 8. Jakob PM, Griswold MA, Edelman RR, Sodickson DK. AUTO-
typically sufficient to achieve good results. However, SMASH: a self-calibrating technique for SMASH imaging. Simul-
taneous acquisition of spatial harmonics. MAGMA 1998;7:42–
especially for coil arrays with many small elements
54.
and high acceleration factors, the resolution of the 9. Heidemann RM, Griswold MA, Haase A, Jakob PM. VD-AUTO-
reference data may need to be increased. SMASH imaging. Magn Reson Med 2001;45:1066 –1074.
Another approach for obtaining the reference data 10. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE:
is an extended variable density (VD) acquisition sensitivity encoding for fast MRI. Magn Reson Med 1999;42:952–
962.
scheme. Here the idea is to acquire the reference data
11. Griswold MA, Jakob PM, Heidemann RM, et al. Generalized auto-
during the accelerated experiment. In a manner calibrating partially parallel acquisitions (GRAPPA). Magn Reson
equivalent to VD-AUTO-SMASH (9), the central k- Med 2002;47:1202–1210.
space is fully sampled while the outer parts of k- 12. Kyriakos WE, Panych LP, Kacher DF, et al. Sensitivity profiles from
space are undersampled along both PE directions. an array of coils for encoding and reconstruction in parallel (SPACE
RIP). Magn Reson Med 2000;44:301–308.
The fully sampled k-space center is extracted to ob-
13. Weiger M, Pruessmann KP, Boesiger P. 2D SENSE for faster 3D
tain the low-resolution reference data, which are MRI. MAGMA 2002;14:10 –19.
used for calibration of the reconstruction process. 14. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The
Furthermore, the reference data may be included in NMR phased array. Magn Reson Med 1990;16:192–225.
450 Blaimer et al.
15. Walsh DO, Gmitro AF, Marcellin MW. Adaptive reconstruction of Annual Meeting of ESMRMB, Copenhagen, Denmark, 2004. Ab-
phased array MR imagery. Magn Reson Med 2000;43:682– 690. stract 398.
16. Wiggins GC, Triantafyllou C, Potthast A, Reykowski A, Nittka M, 18. Griswold MA. Advanced k-space techniques. In: Proceedings of the
Wald LL. A 32 channel receive-only phased array head coil for 3T 2nd International Workshop on Parallel MRI, Zurich, Switzerland,
with novel geodesic tiling geometry. In: Proceedings of the 13th 2004. Abstract 16.
Annual Meeting of ISMRM, Miami Beach, FL, USA, 2005. Abstract 19. Wang Z, Wang J, Detre JA. Improved data reconstruction method
679. for GRAPPA. Magn Reson Med 2005;54:738 –742.
17. Breuer F, Blaimer M, Mueller M, Heidemann RM, Griswold M, 20. Heidemann RM, Griswold MA, Kiefer B, et al. Resolution enhance-
Jakob P. Autocalibrated parallel imaging with GRAPPA using a ment in lung 1H imaging using parallel imaging methods. Magn
single prescan as reference data. In: Proceedings of the 21st Reson Med 2003;49:391–394.