Professional Documents
Culture Documents
Abstract
Parallel magnetic resonance imaging through sensitivity encoding using multiple receiver coils has emerged as an effective tool to reduce
imaging time or to improve image SNR. The quality of reconstructed images is limited by the inaccurate estimation of the sensitivity map,
noise in the acquired k-space data and the ill-conditioned nature of the coefficient matrix. Tikhonov regularization is a popular method to
reduce or eliminate the ill-conditioned nature of the problem. In this approach, selection of the regularization map and the regularization
parameter is very important. Perceptual difference model (PDM) is a quantitative image quality evaluation tool that has been successfully
applied to varieties of MR applications. High correlation between the human rating and PDM score shows that PDM should be suitable to
evaluate image quality in parallel MR imaging. By applying PDM, we compared four methods of selecting the regularization map and four
methods of selecting the regularization parameter. We found that a regularization map obtained using generalized series (GS) together with a
spatially adaptive regularization parameter gave the best reconstructions. PDM was also used as an objective function for optimizing two
important parameters in the spatially adaptive method. We conclude that PDM enables one to do comprehensive experiments and that it is an
effective tool for designing and optimizing reconstruction methods in parallel MR imaging.
D 2006 Elsevier Inc. All rights reserved.
Keywords: Perceptual difference model; Parallel MRI; Image quality; SENSE; Tikhonov regularization
subsampling and slow-acquired images obtained with full features including a measure of contrast [14], and visual de-
sampling, respectively. PDM has been shown to correlate tection threshold [19]. The PDM structure is shown in Fig. 1.
well with human observers in a variety of MR experiments, Inputs to the PDM are an ideal reference image obtained from
including spiral imaging and keyhole imaging [12,13]. In the full-sampled reconstructed image and a test image of one
computer experiments, we use PDM to investigate four of the down-sampled images. Images are windowed to
methods of selecting the regularization parameter and four maximize overall image contrast, and this same windowing
methods of selecting the regularization image. Once we have is maintained during evaluation by human observers de-
identified parameters of interest, we use an optimization scribed later. The regular output of the PDM is a spatial map
algorithm and a PDM objective function to further optimize representing the magnitude of differences that a human
Tikhonov regularization. Noise effects on regularization are observer would perceive between the two images. This map
also investigated. In this study covering multiple image can be summed over a region of interest (ROI), which is
reconstruction methods, continuous reconstruction parame- defined manually to include relevant anatomy, to give a scalar
ters, multiple test images, and variable noise, thousands of PDM error. In this paper, we used an ROI consisting of a
images were generated. It would be impossible for human manually defined ellipse encompassing the whole head area.
observers to review this number of images, and, if they did,
2.2. PDM validation
given the inter- and intra-observer variability, uncertain
results would probably be obtained. The PDM makes such The correspondence between human observer scoring of
experiments possible. image quality and PDM scoring was demonstrated through
experiments performed in a manner similar to that
previously reported by Salem et al. [12] and by Martens
2. Methods and Meesters [20] on a similar model. To test the full range
of image quality in our simulated images, we selected 40
2.1. Perceptual difference model
test images with PDM scores linearly spread between best
The quality of the reconstructed images was evaluated to worst. In the experiment, the 40 images were presented to
with the PDM designed in our laboratory. The model was four subjects, one of the authors and three experienced
inspired by the visible differences predictor model reported radiologists. The image was displayed following gray scale
by Daly [14], but we have made multiple modifications over windowing as reported previously. The region outside of
the years. It has been validated for the evaluation of other fast the region of interest was set to zero value (black). The
MRI applications and described in detail in previous papers evaluation experiment was carried on a Matlab (The
[12,13,15]. It contains components that model the nonline- MathWorks, Natick, MA) GUI program, and all the results
arity in the sensitivity of the retina [16,17], the contrast were automatically recorded. Each presentation consisted of
sensitivity function [16], and the channels of spatial a two-panel display (Fig. 2A), with the high-quality re-
frequency found in the visual cortex [18], as well as other ference image and a randomly selected test image on the left
Fig. 1. A block diagram of the perceptual difference model (PDM). The output is a map showing the likelihood of a perceptual difference between the two input
images. The summation of this map over a region of interest gives a PDM error score. The gold standard image is acquired with a full k-space acquisition and
reconstructed using the total least square method. Subsequent test images are reconstructed from down-sampled k-space data reconstructed with different
regularization methods or regular SENSE method.
D. Huo et al. / Magnetic Resonance Imaging 24 (2006) 123 – 132 125
map. In practice, there are errors in both the estimation of image is increased. We will also consider four methods of
sensitivity maps S and the measured data S. Eq. (1) becomes selecting k:
ðS0 þ DS Þðq0 þ DqÞ ¼ d0 þ Dd ð2Þ ! Set k as a constant over the entire image. This is
the method first proposed for Tikhonov regulariza-
When ||S 0+||||DS|| b1,
tion [22].
Dq j Dd DS ! L-Curve method. The L-curve is a parametric plot of
V þ ð3Þ
q0 1 jODSO=S0 d0 S0 the data fitting error and the deviation from the
regularization image. The idea is that the optimal
where j = ||S 0 ||||S 0+ || is the condition number of S 0 . regularization parameter is to minimize and balance
For nonlocalized coils with large acceleration factors, S 0 the two error terms and it corresponds to the elbow of
tends to be ill-conditioned and leads to a significant recon- the L-curve [25]. This method has been used in
struction error. parallel imaging with some success [10].
2.5. Tikhonov regularization of SENSE reconstruction ! Generalized cross-validation (GCV) [26]. The basic
idea of GCV is that if some data are left out and
Tikhonov regularization is perhaps the most common the solution is computed from the reduced problem,
method to solve the ill-conditioned problem. A weighted then the solution is predicted from the observed data
sum of the data misfit term ||Sqd||2 and a regularization and the choice of the regularization parameter should
term ||qq r||2 with weighting factor k 2 are formed, and the be independent of an orthogonal transformation of
solution is the one that minimizes the sum, the data.
n o ! A spatially adaptive method was proposed [27] to
q4 ¼ arg min OSq dO2 þ k2 Oq qr O2 ð4Þ
q choose k in a spatially dependent fashion, considering
the conditioning of all the equations. Specifically, the
A closed form of the solution could be expressed as
proposed algorithm first sets k to [kmin, kmax] and
1 then selects the k point-by-point in a spatially
q4 ¼ qr þ S H S þ k2 I S H ðd Sqr Þ ð5Þ
adaptive manner. More specifically, k(x) is set to be
The selection of regularization parameter k and regular- a linear function of the local condition number of S;
ization image q r will greatly affect the final reconstructed that is,
image quality.
jðS Þ jmin ðS Þ
kð xÞ¼ ðkmax kmin Þ þ kmin ð6Þ
2.5.1. Selection of regularization image jmax ðS Þ jmin ðS Þ
Researchers have proposed various methods for estimat-
ing the regularization image. We will compare four methods where 8 9
in this paper. >
> >
>
>
< >
=
maxi ri
! Setting q r = 0. This was used in Ref. [22] with kmin ¼ arg min bK
k >
>
> k2 >
>
>
some success. : min i ri þ ;
! Median filter method. The conventional SESNE ri
algorithm is used to obtain an initial reconstruction, ( )
X
which is then filtered by a 55 median filter to kmax ¼ arg max OSqreg ðkÞ dOb E ð7Þ
suppress the residual aliasing artifacts [11]. k x
! Autocalibration method. In this method, additional
k-space center lines at the Nyquist rate is acquired and
Here r i is the ith singular value of sensitivity map
autocalibrated to reconstruct a low-resolution regu-
S. K and E are user-defined constants. In our evaluation
larization image [23].
experiment, they were set as K =10 and E =0.5; they were
! Generalized-series (GS) model method. We will also
further optimized in an optimization experiment.
acquire the additional center k-space lines as we did
in the autocalibration method. An image is then
reconstructed from the k-space data in each coil using 3. Experiments and results
the GS model [24] whose basis functions are formed
3.1. Validation of PDM
from the reference data collected at the Nyquist rate.
Raw data in the PDM validation experiment were
2.5.2. Selection of regularization parameter processed. First, two scores given for the same test image
The choice of the regularization parameter k is also from the same subject were averaged to reduce the intra-
very important to successful regularization. As k increases, observer difference. To compensate for the scale boundary
the contribution to the reconstruction from the data effects, we used a nonlinear scale transformation method
themselves is reduced and that from the regularization recommended by the International Telecommunication
D. Huo et al. / Magnetic Resonance Imaging 24 (2006) 123 – 132 127
Table 1
Correlation coefficient between human subjects, PDM and MSE scores
Subject 1 Subject 2 Subject 3 Subject 4 PDM MSE
Subject 1 0.9707 0.9541 0.9557 0.9626 0.9120 0.4218
Subject 2 0.9541 0.9719 0.9509 0.9529 0.8951 0.4926
Subject 3 0.9557 0.9509 0.9616 0.9738 0.9506 0.5113
Subject 4 0.9626 0.9529 0.9738 0.9855 0.9394 0.4486
PDM 0.9120 0.8951 0.9506 0.9394 1 0.4399
MSE 0.4218 0.4926 0.5113 0.4486 0.4399 1
Correlation coefficients were calculated for all possible combinations of subjects, PDM scores and MSE scores. Note that PDM correlates nearly as well as the
correlation between subjects. The performance of PDM is much better than the performance of MSE.
Union in their report on methods for assessing television Fig. 2(B), the average normalized ratings for each observer
images [28] so as to normalize the human rating data. The and the corresponding PDM error score are plotted as a
mean normalized human observer scores were graphically function of the 40 test images. We could see that the
and statistically compared to the PDM scores on the same performance of PDM is just like a normal human subject.
test images. There is a linear relationship between the human scores
Human observer scoring of image quality was highly averaged over the four subjects and PDM score, as shown in
correlated with PDM scoring of image differences. In Fig. 2(C). The x-intercept (1.13) corresponds to a
Fig. 3. For reduction factor R = 2, PDM score for various q r and k selection combinations. The best solution (lowest PDM score) is shown in (B), corresponding
to the spatially adaptive method to select k and the GS method to select q r. Image reconstructed with full k-space data, or the reference image, is shown
in (C). The image reconstructed with regular SENSE without regularization is shown in (D) and the corresponding PDM score (4.33) is shown as the horizontal
bar in chart (A).
128 D. Huo et al. / Magnetic Resonance Imaging 24 (2006) 123 – 132
b nonperceptible difference.Q Data are further analyzed in Although further investigation is still needed, PDM provides
Table 1. To show the intersubject difference, we use a a possible way to separate the different effects that degrade
simple linear regression model y =mx+b to fit any two the image quality.
subject readings, and the correlation coefficient is shown in
3.2. Comparison of regularization methods
Table 1. The same procedure applies to calculation of the
correlation coefficient between human subject rating and The PDM was first used to select the best regularization
PDM [or mean square error (MSE)]. The intrasubject methods. The test images were reconstructed using reduction
difference is reflected by the correlation coefficient between factors R =2 or 4 (down-sampled by 2 or 4), four methods of
two measurements from the same subject. And all these selecting the regularization image and four methods of
correlation coefficients are listed in Table 1. The PDM selecting the regularization parameter. Images are also
accurately reflects human observer ratings of image quality. reconstructed by the regular SENSE method without regu-
As a comparison, another image quality index, the MSE, larization from the two down-sampled data sets. Image
correlated less well with human observer ratings. quality of all the reconstructed images is evaluated by PDM.
PDM score is not the only possible output format of With R = 2, the impact of the reconstruction methods on
PDM model. One could also get the PDM value in different the image quality is shown in Fig. 3. Four methods of
spatial frequencies. As seen in Fig. 2(A), human scoring of selecting the regularization parameter k and four methods of
noise, aliasing, and other effects was also collected in the selecting regularization image q r generate 16 combinations,
experiments. Our analysis (results not shown here) shows and from the figure we can see that the GS method
that a good correlation exists between the high-frequency combined with the spatially adaptive method gives the best
output of the PDM model and human scoring of the noise solution. The corresponding image is shown in Fig. 3(B).
effects. Accordingly, low-frequency output of the PDM The reference image is shown in Fig. 3(C) and the image
score correlates better with aliasing and other effects. reconstructed with regular SENSE (no regularization) is
Fig. 4. For reduction factor R = 4, PDM score for various q r and k selection combinations. The best solution (lowest PDM score) is shown in (B), corresponding
to the spatially adaptive method to select k and the GS method to select q r. The image reconstructed with regular SENSE without regularization is shown in (C)
and the corresponding PDM score (31.46) is shown as the horizontal bar in chart (A).
D. Huo et al. / Magnetic Resonance Imaging 24 (2006) 123 – 132 129
shown in Fig. 3(D). The PDM score of Fig. 3(D) is shown 3.3. Parameter optimization in the spatially
by the horizontal bar in Fig. 3(A). As expected, down adaptive method
sampling with R = 2 degrades image quality as compared to
From results in the last section, the GS method of
the reference image. As we can see, the unregularized image
selecting q r and the spatially adaptive method of selecting k
gives a PDM score of 4.33, somewhat higher than the
were best. However, the spatially adaptive method needs
bimperceptibleQ threshold. In this case, the application of
two free parameters, K and E. In the previous evaluation, we
regularization helps improve the image quality to some
set them to be 10 and 0.5, based on our previous anecdotal
extent, but the improvement is limited, as observed by the
experience. PDM provides a means of optimizing these free
PDM score and the visual inspection.
parameters. We used a simplex search algorithm that was
Results are shown in Fig. 4 for R = 4. Again, we can see included in Matlab to find the optimal point. During the
that the GS method to select the regularization image optimization process, about 50 images were evaluated, and
combined with the spatially adaptive method offers the best this number should change with different initial conditions
solution. The corresponding image is shown in Fig. 4(B), and stopping criteria. In Fig. 5, we see how the recon-
and the regular SENSE method reconstructed image is structed image quality changes as a function of K and E
shown in Fig. 4(C). In this case, we can see that the regular for R = 4. After searching through the 2-D surface, we can
SENSE reconstruction generates images with significant easily find the optimized values K = 5.91 and E =1.04 cor-
errors when the reduction factor is high, and regularization responding to the lowest PDM score. The optimal image is
improves image quality considerably. shown in Fig. 5(B) corresponding to point A in the image.
Fig. 5. Use PDM score as object function to optimize the free parameters K and E. The surface plot in (A) shows how the PDM score changes with the changes
of K and E. Point A, in which K = 5.91 and E = 1.04, represents the best regularization image [shown in (B)] we can get. Image corresponding to unoptimized
point B, with K = 17 and E = 0.8, is shown in (C).
130 D. Huo et al. / Magnetic Resonance Imaging 24 (2006) 123 – 132
regularization image are another quick and acceptable so- the reduction factor is equal or close to the number of
lution. In the selection of regularization parameter, L-curve coils. The GS method to select the regularization image
and GCV method did not give very good results, and they combined with the spatially adaptive method to select the
are very computationally intensive. We do not recommend regularization parameter gives the best solution for Tikho-
these two methods based on our experiment results. nov regularization.
It is exciting that we can use the PDM to optimize free
parameters in the reconstruction. Optimization is advisable,
Acknowledgments
as shown by the large difference between the optimized and
unoptimized images in Fig. 5. The surface in Fig. 5(A) is This work was supported under NIH grant R01
very smooth, indicating that one need not be overly EB004070 and NSF grant NSF-BES-0201876. We thank
concerned with local minima. Indeed, we tested several for the contribution from Leslie Ying in University of
different initial values, and each converged to the same Wisconsin, Milwaukee, WI, and the subjects for participating
result. In addition, as described previously, the optimum was in the experiments.
reasonably insensitive to added noise.
In the reconstruction optimization process, we had to
References
define a stopping criterion. Experimentally, we determined
that a tolerance of 1% gave nearly optimal results and saved [1] Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The
NMR phased-array. Magn Reson Med 1990;16:192 – 225.
considerable time as compared to a much tighter tolerance.
[2] Hutchinson M, Raff U. Fast MRI data acquisition using multiple
PDM allows the MR-imaging engineer to make quanti- detectors. Magn Reson Med 1988;6:87 – 91.
tative trade-offs, and the experiments with added noise give [3] Kwiat D, Einav S, Navon G. A decoupled coil detector array for fast
an excellent example. We can clearly see the effect of added image acquisition in magnetic-resonance-imaging. Med Phys 1991;
noise on image quality in Fig. 6. Image quality is improved 18:251 – 65.
[4] Ra JB, Rim CY. Fast imaging using subencoding data sets from
with increasing SNR of acquired data, but the improvement
multiple detectors. Magn Reson Med 1993;30:142 – 5.
is limited above 50 dB because the image quality is now [5] Sodickson DK, Manning WJ. Simultaneous acquisition of spatial
limited by the inaccurate estimation of the sensitivity map harmonics (SMASH): fast imaging with radiofrequency coil arrays.
and the ill condition of the inverse problem. The 50- dB Magn Reson Med 1997;38:591 – 603.
value thus provides a guideline; one need not design a [6] Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE:
sensitivity encoding for fast MRI. Magn Reson Med 1999;42:
system with SNR exceeding much beyond this, possibly
952 – 62.
saving acquisition time. We can also see the importance of [7] Griswold MA, Jakob PM, Heidemann RM, Nittka M, Jellus V, Wang
accurate estimation of the sensitivity map. Often, low- JM, et al. Generalized Autocalibrating Partially Parallel Acquisitions
resolution, fast sequences are used to get the sensitivity (GRAPPA). Magn Reson Med 2002;47:1202 – 10.
map. From Fig. 6, prescan SNR can degrade image quality [8] de Zwart JA, Ledden PJ, Kellman P, van Gelderen P, Duyn JH. Design
of a SENSE-optimized high-sensitivity MRI receive coil for brain
below 30 dB. One can also compare image quality with
imaging. Magn Reson Med 2002;47:1218 – 27.
regularization (Fig. 6) to normal SENSE reconstruction [9] Weiger M, Pruessmann KP, Leussler C, Roschmann P, Boesiger P.
(Fig. 4), for R =4. Even at the lowest SNR (20 dB for Specific coil design for SENSE: a six-element cardiac array. Magn
both acquisition and prescan), image quality following Reson Med 2001;45:495 – 504.
regularization exceeded that for regular SENSE with no [10] Lin FH, Kwong KK, Belliveau JW, Wald LL. Parallel imaging re-
construction using automatic regularization. Magn Reson Med 2004;
added noise. This indicates the power of improved
51:559 – 67.
reconstruction techniques. [11] Liang ZP, Bammer R, Ji J, Pelc N, Glover G. Making better SENSE:
Although results presented here were generated from a wavelet denoising, Tikhnov regularization, and total least squares.
single MR data set, we have spot tested results on other Proceedings of 10th Annual Meeting of ISMRM, 2002.
brain data sets having different acquisition sequences and [12] Salem KA, Lewin JS, Aschoff AJ, Duerk JL, Wilson DL. Validation
of a human vision model for image quality evaluation of fast
different methods to estimate the sensitivity map, and results
interventional magnetic resonance imaging. J Electron Imaging 2002;
are similar. The best regularization method is still the GS 11:224 – 35.
method to select q r and the spatially adaptive method to [13] Wilson DL, Salem KA, Huo D, Duerk JL. Perception difference
select k. Furthermore, using K = 5.91 and E = 1.09 generated paradigm for analyzing image quality of fast MRI techniques.
near optimal PDM score (within 1% difference) in other Proceedings of SPIE Medical Imaging on Image Perception, Observer
Performance, and Technology, 2003.
cases. Therefore, we can reasonably assume that results are
[14] Daly S. The visual differences predictor: an algorithm for the
generalizable to some other conditions. assessment of image fidelity. In: Watson AB, editor. Digital images
and human vision. Cambridge (MA)7 MIT Press; 1993. p. 179 – 206.
[15] Flask CA, Salem KA, Lewin JS, Duerk JL. Keyhole Dixon method for
5. Conclusion faster fat suppression with perceptual equivalence. Radiology 2002;
225:314.
PDM has been successfully applied to evaluate the
[16] Mannos JL, Sakrison DJ. The effects of a visual fidelity criterion on
regularization methods in parallel MR imaging and to the encoding of images. IEEE Trans Info Theory 1974;IT-20:525 – 36.
optimize the important free parameters. Tikhonov regular- [17] McCann JJ, McKee SP, Taylor TH. Quantitative studies in retinex
ization is very important to improve the image quality when theory: a comparison between theoretical predictions and observer
132 D. Huo et al. / Magnetic Resonance Imaging 24 (2006) 123 – 132
responses to the bColor MondrianQ experiments. Vision Res 1976; Multaneous Acquisition of Spatial Harmonics. MAGMA 1998;7:
16:445 – 58. 42 – 54.
[18] Watson AB. The cortex transform — rapid computation of simu- [24] Liang ZP, Stenger A, Ji JX, Ma J, Boada F. Parallel generalized series
lated neural images. Comput Vis Graph Image Process 1987;39: imaging. Proceedings of 11th Annual Meeting of ISMRM, 2003.
311 – 27. [25] Hansen PC. Analysis of discrete ill-posed problems by means of the
[19] Normann RA, Baxter BS, Ravindra H, Anderton PJ. Photoreceptor L-curve. Siam Rev 1992;34:561 – 80.
contributions to contrast sensitivity — applications in radiological- [26] Wahba G, Wang YH. When is the optimal regularization parameter
diagnosis. IEEE Trans Syst Man Cybern 1983;13:944 – 53. insensitive to the choice of the loss function. Commun Stat Theory
[20] Martens JB, Meesters L. Image dissimilarity. Signal Process 1998; Methods 1990;19:1685 – 700.
70:155 – 76. [27] Ying L, Xu D, Liang ZP. On Tikhonov regularization for image
[21] Macovski A. Noise in MRI. Magn Reson Med 1996;36:494 – 7. reconstruction in parallel MRI. Proceedings of 26th Annual Interna-
[22] King KF, Angelos L. SENSE image quality improvement using tional Conference of Engineering in Medicine and Biology Society,
matrix regularization. Proceedings of 9th Annual Meeting of 2004;1:1056 – 9.
ISMRM, 2001. [28] International Telecommunication Union. ITU-R BT.500-11 Method-
[23] Jakob PM, Griswold MA, Edelman RR, Sodickson DK. AUTO- ology for the subjective assessment of the quality of television
SMASH: a self-calibrating technique for SMASH imaging. Si- pictures; 2002. Available at http://www.itu.int.