Professional Documents
Culture Documents
State of The Art Iterative Reconstrution
State of The Art Iterative Reconstrution
incorporating an accurate statistical modeling of the measure- by Xu et al. [34], which was derived with the consideration of
ment is a prerequisite of the SIR algorithms [11]–[13], [18] and sparse representation theory via a sparse linear combination of
the edge-preserving regularization term plays an important role the patch-based atoms in the dictionary.
in the successful image reconstruction. Usually, the regulariza- In this paper, we aim to lower mAs for radiation reduction
tion term is chosen as a shift-invariant function that penalizes the in the case of repeated CT scans and propose a prior-image
differences among local neighboring pixels. These regulariza- induced nonlocal (PINL) regularization for SIR via the penal-
tions through equally smoothing both the noise and edge details ized weighted least-squares (PWLS) criteria, which we refer
often tend to produce an unfavorable oversmoothing effect [12]. to as “PWLS-PINL.” Specifically, the present PINL regular-
In contrast to the smooth regularization, many edge-preserving ization utilizes the redundant information in the prior-image
regularizations/priors were proposed in the literature [11], [22]. and the weighted least-squares term considers a data-dependent
A typical example is the Huber prior [19]–[21], which replaces variance estimation, aiming to improve current low-dose image
the quadratic penalty function with a nonquadratic penalty func- quality. Subsequently, a modified iterative successive overrela-
tion that increases less rapidly compared with the quadratic tion algorithm is adopted to optimize the associative objective
penalty function for sufficiently large arguments. However, function.
these edge-preserving regularizations/priors mostly rely on the
properties of the local smoothness or edges and do not con- II. METHODS AND MATERIALS
sider the basic affinity structure information of the desired im-
A. PWLS Criteria for CT Image Reconstruction
age, such as the gray levels, edge indicator, dominant direction,
and dominant frequency. To address the aforementioned issues Mathematically, the X-ray CT measurement can be approxi-
of the conventional regularizations/priors, an edge-preserving mately expressed as a discrete linear system
nonlocal (NL) prior was proposed for CT and positron emis-
y = Hμ (1)
sion tomography image reconstructions with encouraging re-
sults [22]–[24] by fully exploiting the density difference infor- where μ denotes the vector of attenuation coefficients, i.e.,
mation and the nonlocal connectivity and continuity information μ = (μ1 , μ2 , . . . , μN ) , and y represents the obtained sinogram
of the desired image. data (projections after system calibration and logarithm trans-
With regard to the repeated CT scans, a previously scanned formation), i.e., y = (y1 , y2 , . . . , yM ) , where “ ” denotes the
high-quality diagnostic CT image volume usually contains same matrix transpose. The operator H represents the system or pro-
anatomical information with the current scan except for some jection matrix with a M × N size. The element of Hij denotes
anatomical changes due to internal motion or patient weight the length of intersection of projection ray i with pixel j. In
change. Generally, the CT scans at different times are often our implementation, the associated element was precalculated
dealt independently and no systematic attempt has been made by a fast ray-tracing technique [32] and stored as a file. The
to integrate the valuable patient-specific prior knowledge, i.e., goal of CT image reconstruction is to estimate the attenuation
the previous scanned data that hold a wealth of prior informa- coefficients μ from the measurement y.
tion on the patient-specific anatomy, to promote the subsequent According to the measurement model (1) and the MAP esti-
imaging process. Undoubtedly, performing a low-dose proto- mation criterion [12], [33], the mathematical formula for PWLS
col in the follow-up CT scan by fully using the previous image image reconstruction with a regularization term R(μ) can be ex-
into the current image reconstruction framework, is a promising pressed as
and interesting research topic [25], [28], [29], [31], [34]. For
μ∗ = arg min (y − Hμ) Σ−1 (y − Hμ) + βR(μ) (2)
instance, Chen et al. proposed a prior image constrained com- μ≥0
pressed sensing (PICCS) algorithm [29] to enable view angle
undersampling by integrating a prior image into the reconstruc- where Σ is a diagonal matrix with the ith element of σi2 which
tion procedure. Lauzier extended the PICCS algorithm to the is the variance of sinogram data y. β is a hypeparameter that
DR-PICCS algorithm [30] for CT radiation dose reduction us- controls the strength of regularization.
ing a statistical model. A weakness of PICCS is that the prior and To determine the parameter σi2 in (2), several methods can be
current images are taken at the same global geometrical coordi- used [12], [34]. In this study, the variance of σi2 was determined
nates. This assumption, however, does not necessarily translate by the mean-variance relationship proposed by Ma et al. [27],
into practical settings like in the IGRT applications. Accurate which is written as
registration and voxel consistency may limit the wide use of the 1 1
2
σi (y) = exp(p̄i ) 1 + exp(p̄i )(σe − 1.25)
2
(3)
PICCS algorithm. To address this issue, Ma et al. [10] proposed I0 I0
a low-dose CT image filtering method (named as “ndiNLM” al-
where I0 denotes the incident X-ray intensity, p̄i is the mean of
gorithm) by utilizing a high-quality normal-dose scan as priori
sinogram data y at bin i, and σe2 is the background electronic
information to perform current low-dose CT image restoration
noise variance.
based on nonlocal means criteria [26]. The ndiNLM algorithm
performed well in noise reduction, but actually it is a postpro-
cessing approach without considering the statistical property of B. Overview of the PINL regularization
the CT projection data. Another strategy that relaxes image reg- Traditionally, R(μ) in (2) is designed by a simple weighted
istration was the dictionary learning-based approach proposed sum of the potential function on the different values of the
ZHANG et al.: ITERATIVE RECONSTRUCTION FOR X-RAY COMPUTED TOMOGRAPHY 2369
neighboring pixels in the image domain [12] and can be de- the registered prior image from the initial high-dose CT im-
scribed as age. The effectiveness of the B-spline-based image registration
algorithm has been extensively validated by many registration
R(μ) = R(μj ) = w(k, j)ϕ(μj − μk ) (4) experiments. Based on the transformation matrix, all voxels of
j j k ∈S j
the prior CT images are then transformed into the current CT
where index j runs over all image elements in the image do- images to obtain globally aligned prior images. Similar to our
main, Sj represents the local neighborhood of the jth image previous studies in low-dose CT image restoration [10], through
pixel in two dimensions, and ϕ denotes a convex and positive two roughly aligned images, the rich redundant information in
potential function satisfying ϕ(0) = 0. Weight w(k, j) is pos- the prior image can be effectively used to induce the PINL regu-
itive and symmetric, i.e., w(k, j) ≥ 0 and w(k, j) = w(j, k), larization for low-dose image reconstruction. Further evaluation
which are usually designed as an inverse proportion of the dis- of image registration in our study is presented in Section III-A5.
tance between the pixels k and j in Sj . These regularizations After we obtained the registered prior image μreg prior , weight
are inherently local because they lack global connectivity or w(k, j) in (6) can be regarded as a function of the current es-
continuity. timation μ and the registered prior image μreg prior . Because of
In the repeated CT scans, a previously scanned normal-dose the nonlinearity in calculating weight w(k, j), minimizing the
CT image has lower noise level and higher resolution than the objective function of (9) is difficult for a closed solution. To
current low-dose CT image, and majority of the anatomical in- address this problem, based on our previous studies [23], [28],
formation are same between the two scans. Thus, in this paper, a binary optimal scheme can be used to minimize the objective
to utilize fully the redundancy of information in the prior image, function of (9). Specifically, weight w(k, j) is automatically ad-
we propose a prior-image induced nonlocal (PINL) regulariza- justed according to the similarity between the patch-windows
tion with quadratic potential function, which can be written as in the current estimation μn (n is the iterative index) and the
registered prior image μreg
prior during each iteration.
RPINL (μ) = w(k, j)(μj − μreg
prior,k )
2
(5) In this study, we modified the iterative successive overre-
j k ∈N j laxation algorithm described in [12] and [38] to calculate the
solution of (9). In the implementation, the variance σi2 , i.e., the
where μreg
prior denotes the registered prior image between the element of weight matrix in (9), is updated in each iteration
prior image μprior and the current low-dose image μ, Nj denotes according to (3) for a more accurate estimation of the sino-
the search window in the image μreg prior . The nonlocal weight gram variance. For simplicity, we call our implementation as
w(k, j) between the prior image and the current objective image the PWLS-PINL algorithm, and similar to [12], [38], it can be
is defined as summarized as follows:
w(k, j) = exp(−D(k, j)/h2 )/ exp(−D(k, j)/h2 ) (6)
k ∈N j 1 Initialization
2 μ= FBP{y}
D(k, j) = μ(nj ) − μreg 2
prior (nk )2 (7)
3 r̂=y−Hμ̂
μ(nj ) = {μl : l ∈ nj }, μreg reg
prior (nk ) = {μprior,l : l ∈ nk }. (8) 4 Σ= diag{σi2 (yi )}
Here, weight w(k, j) is a decreasing function of the similarity 5 Sj = H j Σ−1 Hj , ∀ j.
between two local neighborhoods nk and nj (named patch- 6 For n = 1, 2, 3, . . . , IterMax,
window) centered at the pixel k in the image domain of μreg prior
7 Update {w(k, j)} according to Eq. (6).
and the pixel j in the image domain of μ, respectively. ·22
represents the 2-D Euclidean distance between two similar- 8 For each voxel,
ity patch-windows. μ(nj ) and μreg prior (nk ) denote the vector of
9 μ̂old
j := μ̂j
neighborhood pixel values restricted in the patch-windows nj 10 αj := Sj + β k ∈N j w(k, j), ∀ j
and nk , respectively, and h is a parameter that controls the decay
H j Σ −1 r̂ +S j μ̂ oj l d +β k ∈N j w (k ,j )μ rpergi o r , k
of the exponential function. 11 μ̂new
j := αj
In summary, the objective function of the PWLS with the
12 μ̂j := max 0, (1 − ω)μold
j + ω μ̂new
j
present PINL regularization (i.e., PWLS-PINL) for CT image
reconstruction can be rewritten as 13 r̂ := r̂ + Hj (μ̂old
j − μ̂j ).
14 End
μ∗ = arg min (y − Hμ) Σ−1 (y − Hμ) + βRPINL (μ) . (9)
μ≥0 16 Σ := diag{σi2 ( j Hij μ̂j )}.
17 End
C. Implementation of the PWLS-PINL Method
In the implementation of the present PWLS-PINL method, where IterMax denotes the total iterative number, the parameter
the current CT images and the initial high-dose CT images are ω should satisfy 0 ≤ ω ≤1. The iterative process is terminated
first reconstructed by the FBP method, and then a B-spline- if certain convergence criteria is satisfied for a relatively stable
based image registration technique [37] is adopted to generate solution [12], [22]. In this study, the 200th iterated image was
2370 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 61, NO. 9, SEPTEMBER 2014
0.5 s, the slice thickness was 5.0 mm, the tube voltage was
120 kVp, the tube currents are 40 and 400 mA.
E. Evaluation Merits
1) Evaluation by Noise Reduction: The following metrics
were utilized to evaluate the noise reduction on the reconstructed
image from the low-dose sinogram data: 1) signal to noise ratio
(SNR); 2) local signal to noise ratio (lSNR); and 3) relative root
mean square error (rRMSE)
Q
m =1 (μm − μ̄)
2
SNR = 10 log10 Q (10)
m =1 (μm − μGS,m )
2
Fig. 1. Illustration of an anthropomorphic torso phantom (a), and (b) a CT
image reconstructed by the FBP method with an optimized Hamming filter from
1
Q
the sinogram data acquired with 100 mAs, 120 kVp. Q m =1 μm
lSNR =
2 (11)
Q Q
m =1 μm − Q
1 1
selected as the result which corresponds to a stable estimation Q m =1 μm
F. Comparison Methods
To validate and evaluate the performance of the present
PWLS-PINL method, the FBP method using ramp filter, the
NL-prior-based PWLS (PWLS-NL) method [23], [24], the
PICCS method [29], [30] and a global dictionary-based sta-
tistical iterative reconstruction (GDSIR) approach [34] were
adopted for comparison. The NL-prior term in the PWLS-NL Fig. 2. Anthropomorphic torso phantom reconstructions by using different
approach is expressed as methods from the sinogram dataset acquired with 17 mAs, 120 kVp. (a) FBP
reconstruction with ramp filter; (b) PWLS-NL reconstruction (N = 21 × 21,
n = 5 × 5, h = 2.0 × 10−3 , β = 2.5 × 104 ); (c) PICCS reconstruction (α =
RNL (μ) = w(k, j)(μj − μk )2 (18) 0.6, λ = 0.1); (d) image reconstructed by the GDSIR method (γ = 4 × 10−2 )
j k ∈N j with a global dictionary trained from the prior image of Fig. 1(b); and
(e) PWLS-PINL reconstruction using Fig. 1(b) as the prior image (N =
where in weight w(k, j) is defined as 21 × 21, n = 5 × 5, h = 2.0 × 10−3 , β = 2.0 × 104 ). All the images are dis-
played in a same window.
w(k, j) = exp − μ(nj ) − μ(nk )22 /h2 (19)
μ(nj ) = {μl : l ∈ nj }, μ(nk ) = {μ : l ∈ nk } (20) where D ∈ RN ×K represents the trained dictionary whose col-
umn dk ∈ RN ×1 (k = 1, . . . , K) is a N dimensional vector,
where nj and nk represent the patch-windows, and Nj denotes
αs ∈ RK ×1 is the representation vector with few nonzero en-
the search window in the current low-dose CT image domain.
tries, and · 0 denotes the l0 -norm. Es ∈ RK ×K is the matrix
The PICCS algorithm under the PWLS framework can be
to extract a patch from the image μ. vs is the Lagrange mul-
formulated as [30]:
tiplier and γ is a hyperparameter. In this study, the global dic-
μ∗ = arg min[αΨ1 (μ − μprior )1 + (1 − α)Ψ2 (μ)1 ] tionary with the size of 64 × 256, (N = 8 × 8, K = 256) was
μ≥0
generated from a normal-dose prior image by using a K-SVD
λ algorithm [35] wherein an orthogonal matching pursuit tech-
+ (y − Hμ) Σ−1 (y − Hμ). (21)
2 nique [36] was used for sparse representation. Additionally, an
where μprior is the prior image, Ψ1 and Ψ2 are the sparsifying alternating minimization scheme described in [34] was adapted
transforms, and α ∈ [0, 1] is a scalar that controls the relative to find the solution of (22).
weights to be allocated to the prior image and compressing
sensing terms. x1 represents the l1 -norm defined as x1 =
III. RESULTS
i |xi |. In the implementation, the sparsifying transform used
A. Anthropomorphic Torso Phantom Study
for both Ψ1 and Ψ2 is the image spatial gradient norm, which is
also equivalent to the total variation (TV). 1) Noise and Artifacts Suppression: Figs. 2 and 3 show the
The GDSIR is a patch-based approach by extracting the prior results reconstructed by five different methods from the sino-
information via a global dictionary trained from a high-quality gram data acquired at 17 and 40 mAs, respectively. Figs. 2(a)
image [34], and the associative image reconstruction is equiva- and 3(a) show the images reconstructed by the FBP method
lent to solve the following minimization problem: with ramp filter. Serious noise-induced streak artifacts can be
observed. Figs. 2(b) and 3(b) show the images reconstructed by
{μ∗ , α∗ } = arg min (y − Hμ) Σ−1 (y − Hμ)
μ,α the PWLS-NL method. The noise was mostly suppressed, but
undesired streak artifacts can still be observed in the highly at-
+γ Es μ − Dαs 22 + vs αs 0 (22) tenuated regions, partially because of the properties of nonlocal
s s weight calculations for structuring preserving. Figs. 2(c) and
2372 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 61, NO. 9, SEPTEMBER 2014
(a)
TABLE III
Fig. 5. Zoomed details of the four ROIs in Fig. 2. CNR MEASURES OF THE IMAGES RECONSTRUCTED BY THE FOUR DIFFERENT
METHODS FROM THE SINOGRAM DATA AT 17 AND 40 MAS
(a) (b)
Fig. 9. Global rRMSE and SNR measures of the present PWLS-PINL method
with respect to the number of iterations from the anthropomorphic torso phantom
reconstructions in the two cases of 17 mAs and 40 mAs. (a) rRMSE versus the
number of iterations; (b) SNR versus the number of iterations.
Fig. 11. Influence of the mismatch between the first prior image and the
desired image on the performance of the PWLS-PINL and PICCS methods.
(a) the FBP image reconstructed with ramp filter from the low-dose sinogram
data; (b) unregistered prior image reconstructed from the normal dose sinogram
data using the FBP method with ramp filter; (c) image reconstructed by the
PICCS method from the low-dose sinogram data using (b) as the prior-image
(α = 0.7, λ = 8.0 × 10−3 ); and (d) image reconstructed by the PWLS-PINL
method from the low-dose sinogram data using (b) as the prior image (N = 35 ×
35, n = 5 × 5, h = 1.0 × 10−3 , β = 5.0 × 105 ). The zoomed ROIs indicated
by three squares in (c) and (d) are displayed at the bottom, respectively. All the
images are displayed in a same window.
Fig. 10. Influence of registration accuracy on the performance of the
PWLS-PINL method. (a) Elastic deformed FBP image from the original one
reconstructed from the sinogram data acquired at 100 mAs, 120 kVp; (b) Corre-
sponding registered image by using the B-spline-based registration algorithm;
(c) image reconstructed by the PWLS-PINL method from the sinogram data ac-
quired at 17 mAs, 120 kVp using (a) as the prior-image; (d) image reconstructed
by the PWLS-PINL method using (b) as the prior-image from the sinogram data
acquired at 17 mAs, 120 kVp; (e) difference image between (c) and Fig. 2(a);
and (f) difference image between (d) and Fig. 2(a). All the images are displayed
in a same window.
a global optimization. Similar to many existing OSL algorithms dose X-ray computed tomography,” IEEE Trans. Med. Imag., vol. 25,
where global convergence is an open issue, the present algo- no. 10, pp. 1272–1283, Oct. 2006.
[13] S. Singh, M. K. Kalra, M. D. Gilman, J. Hsieh, H. H. Pien,
rithm also suffered from the lack of strict global convergence S. R. Digumarthy, and J. A. Shepard, “Adaptive statistical iterative re-
proof. Despite the lack of guarantee of global convergence, the construction technique for radiation dose reduction in chest CT: A pilot
present algorithm is still effective in practice together using the study,” Radiology, vol. 259, no. 2, pp. 565–573, May 2011.
[14] L. Ouyang, T. Solberg, and J. Wang, “Effects of the penalty on the penal-
joint optimization scheme. ized weighted least-squares image reconstruction for low-dose CBCT,”
Another major drawback of the present PWLS-PINL algo- Phys. Med. Biol., vol. 56, no. 17, pp. 5535–5552, Sep. 2011.
rithm is that the update of the nonlocal weights matrix unavoid- [15] J. P. Guedon and Y. Bizais, “Bandlimited and Haar filtered back-projection
reconstructions,” IEEE Trans. Med. Imag., vol. 13, no. 3, pp. 430–440,
ably increases the computational burden as described in [28]. Sep. 1994.
To address this issue, one possible strategy is to use infrequent [16] H. G. Hofmann, B. Keck, C. Rohkohl, and J. Hornegger, “Comparing
update of the parameters (nonlocal weights matrix) interleaved performance of many-core CPUs and GPUs for static and motion com-
pensated reconstruction of C-arm CT data,” Med. Phys., vol. 38, no. 1,
with more frequent OSL steps; and another one is to use fast pp. 468–473, Jan. 2011.
computers and dedicated hardware [45] to develop parallel it- [17] X. Jia, B. Dong, Y. Lou, and S. B. Jiang, “GPU-based iterative cone-beam
erative image reconstruction algorithm. We believe that in the CT reconstruction using tight frame regularization,” Phys. Med. Biol.,
vol. 56, no. 13, pp. 3787–3807, Jul. 2011.
future, most iterative-based image reconstruction algorithms in- [18] L. M. Mitsumori, W. P. Shuman, J. M. Busey, O. Kolokythas, and
cluding the present PWLS-PINL may be widely used in medical K. M. Koprowicz, “Adaptive statistical iterative reconstruction versus fil-
clinic. tered back projection in the same patient: 64 channel liver CT image quality
and patient radiation dose,” Eur. Radiol., vol. 22, no. 1, pp. 138–143, Jan.
2012.
ACKNOWLEDGMENT [19] S. Z. Li, Markov Random Field Modeling in Image Analysis. Tokyo,
Japan: Springer-Verlag, 2001.
The authors would like to thank the anonymous reviewers [20] D. Shulman and J Herve, “Regularization of discontinuous flow fields,”
for their constructive comments and suggestions that greatly in Proc. Workshop Vis. Motion, Irvine, CA, USA, 1989, pp. 81–86.
[21] M. Black and G. Sapiro, “Edges as outliers: Anisotropic smoothing using
improve the quality of the manuscript. local image statistics,” in Proc. Scale-Space Theories Comput. Vis., Corfu,
Greece, 1999, pp. 259–270.
REFERENCES [22] D. F. Yu and J. A. Fessler, “Edge-preserving tomographic reconstruction
with nonlocal regularization,” IEEE Trans. Med. Imag., vol. 21, no. 2,
[1] T. Schubert, A. L. Jacob, M. Takes, T. Menter, A. Gutzeit, and S. Kos, pp. 159–173, Feb. 2002.
“CT-guided percutaneous biopsy of a mass lesion in the upper presacral [23] J. Ma, Q. Feng, Y. Feng, J. Huang, and W. Chen, “Generalized Gibbs priors
space: A sacral transneuroforaminal approach,” Cardiovasc. Intervent. based positron emission tomography reconstruction,” Comput. Biol. Med.,
Radiol., vol. 35, no. 5, pp. 1255–1257, Oct. 2012. vol. 40, no. 6, pp. 565–571, Jun. 2010.
[2] J. H. Siewerdsen, “Cone-beam CT with a flat-panel detector: From image [24] Y. Chen, D. Gao, C. Nie, L. Luo, W. Chen, X. Yin, and Y. Lin, “Bayesian
science to image-guided surgery,” Nucl. Instrum. Methods Phys. Res. A, statistical reconstruction for low-dose X-ray computed tomography us-
vol. 648, no. S1, pp. S241–S250, Aug. 2011. ing an adaptive-weighting nonlocal prior,” Comput. Med. Imag. Graph.,
[3] S. B. Dharap, A. A. Khandkar, A. Pandey, and A. K. Sharma, “Repeat CT vol. 33, no. 7, pp. 495–500, Oct. 2009.
scan in closed head injury,” Injury, vol. 36, no. 3, pp. 412–416, Mar. 2005. [25] H. Yu, S. Zhao, E. A. Hoffman, and G. Wang, “Ultra-low dose lung CT
[4] B. Zhao, L. P. James, C. S. Moskowitz, P. Guo, M. S. Ginsberg, perfusion regularized by a previous scan,” Acad. Radiol., vol. 16, no. 3,
R. A. Lefkowitz, Y. Qin, G. J. Riely, M. G. Kris, and L. H. Schwartz, pp. 363–373, Mar. 2009.
“Evaluating variability in tumor measurements from same-day repeat CT [26] A. Buades, B. Coll, and J. M. Morel, “A nonlocal algorithm for image
scans of patients with non-small cell lung cancer,” Radiology, vol. 252, denoising,” in Proc. IEEE Comput. Vis. Pattern Recog, 2005, vol. 2,
no. 1, pp. 263–272, Jul. 2009. pp. 60–65.
[5] Y. C. Kuo, T. H. Wu, T. S. Chung, K. W. Huang, K. S. Chao, W. C. Su, [27] J. Ma, Z. Liang, Y. Fan, Y. Liu, J. Huang, W. Chen, and H. Lu, “Vari-
and J. F. Chiou, “Effect of regression of enlarged neck lymph nodes on ance analysis of x-ray CT sinograms in the presence of electronic noise
radiation doses received by parotid glands during intensity-modulated background,” Med. Phys., vol. 39, no. 7, pp. 4051–4065, Jul. 2012.
radiotherapy for head and neck cancer,” Amer. J. Clin. Oncol., vol. 29, [28] J. Ma, H. Zhang, Y. Gao, J. Huang, Z. Liang, Q. Feng, and W. Chen,
no. 6, pp. 600–605, Dec. 2006. “Iterative image reconstruction for cerebral perfusion CT using a pre-
[6] L. Yu, X. Liu, S. Leng, J. M. Kofler, J. C. Ramirez-Giraldo, M. Qu, contrast scan induced edge-preserving prior,” Phys. Med. Biol., vol. 57,
J. Christner, J. G. Fletcher, and C. H. McCollough, “Radiation dose reduc- no. 22, pp. 7519–7542, Nov. 2012.
tion in computed tomography: Techniques and future perspective,” Imag. [29] G. H. Chen, P. Theriault-Lauzier, J. Tang, B. Nett, S. Leng, J. Zambelli,
Med., vol. 1, no. 1, pp. 65–84, Oct. 2009. Z. Qi, N. Bevins, A. Raval, S. Reeder, and H. Rowley, “Time-resolved
[7] C. H. McCollough, M. R. Bruesewitz, and J. M. Kofler, Jr., “CT dose interventional cardiac C-arm cone-beam CT: An application of the PICCS
reduction and dose management tools: Overview of available options,” algorithm,” IEEE Trans. Med. Imag., vol. 31, no. 4, pp. 907–923, Apr.
Radiographics, vol. 26, no. 2, pp. 503–512, Mar./Apr. 2006. 2012.
[8] M. K. Kalra, M. M. Maher, T. L. Toth, B. Schmidt, B. L. Westerman, [30] P. T. Lauzier and G. H. Chen, “Characterization of statistical prior image
H. T. Morgan, and S. Saini, “Techniques and applications of automatic constrained compressed sensing (PICCS)—Part II. Application to dose
tube current modulation for CT,” Radiology, vol. 233, no. 3, pp. 649–657, reduction,” Med. Phys., vol. 40, no. 2, pp. 021902:1–021902:14, Feb.
Dec. 2004. 2013.
[9] L. Yu, M. R. Bruesewitz, K. B. Thomas, J. G. Fletcher, J. M. Kofler, and [31] L. Ouyang, T. Solberg, and J. Wang, “Noise reduction in low-dose cone
C. H. McCollough, “Optimal tube potential for radiation dose reduction beam CT by incorporating prior volumetric image information,” Med.
in pediatric CT: Principles, clinical implementations, and pitfalls,” Radio- Phys., vol. 39, no. 5, pp. 2569–2577, May 2012.
graphics, vol. 31, no. 3, pp. 835–848, May/Jun. 2011. [32] G. Han, Z. Liang, and J. You, “A fast ray-tracing technique for TCT and
[10] J. Ma, J. Huang, Q. Feng, H. Zhang, H. Lu, Z. Liang, and W. Chen, “Low- ECT studies,” in Proc. IEEE Nucl. Sci. Symp., Seattle, WA, USA, 1999,
dose computed tomography image restoration using previous normal-dose vol. 3, pp. 1515–1518.
scan,” Med. Phys., vol. 38, no. 10, pp. 5713–5731, Oct. 2011. [33] T. Li, X. Li, J. Wang, J. Wen, H. Lu, J. Hsieh, and Z. Liang, “Nonlinear
[11] J. Wang, T. Li, and L. Xing, “Iterative image reconstruction for CBCT sinogram smoothing for low-dose X-ray CT,” IEEE Trans. Nucl. Sci.,
using edge-preserving prior,” Med. Phys., vol. 36, no. 1, pp. 252–260, Jan. vol. 51, no. 5, pp. 2505–2513, Oct. 2004.
2009. [34] Q. Xu, H. Yu, X. Mou, L. Zhang, J. Hsieh, and G. Wang, “Low-dose X-
[12] J. Wang, T. Li, H. Lu, and Z. Liang, “Penalized weighted least-squares ray CT reconstruction via dictionary learning,” IEEE Trans. Med. Imag.,
approach to sinogram noise reduction and image reconstruction for low- vol. 31, no. 9, pp. 1682–1697, Sep. 2012.
2378 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 61, NO. 9, SEPTEMBER 2014
[35] M. Elad and M. Aharon, “Image denoising via sparse and redundant [41] M. J. Black, G. Sapiro, D. H. Marimont, and D. Heeger, “Robust
representations over learned dictionaries,” IEEE Trans. Image Process., anisotropic diffusion,” IEEE Trans. Image Process., vol. 7, no. 3, pp. 421–
vol. 15, no. 12, pp. 3736–3745, Dec. 2006. 432, Mar. 1998.
[36] Y. C. Pati, R. Rezaiifar, and P. Krishnaprasad, “Orthogonal matching pur- [42] O. Dandekar, K. Siddiqui, V. Walimbe, and R. Shekhar, “Image registra-
suit: Recursive function approximation with applications to wavelet de- tion accuracy with low-dose CT: How low can we go?” in Proc. IEEE Int.
composition,” in Proc. 27th Asilomar Conf. Signals Syst. Comput., Pacific Symp. Biomed. Imag., Arlington, VA, USA, 2006, pp. 502–505.
Grove, CA, USA, 1993, vol. 1, pp. 40–44. [43] D. V. De Ville and M. Kocher, “SURE-based non-local means,” IEEE
[37] R. Szeliski and J. Coughlan, “Spline-Based Image Registration,” Int. J. Signal Process. Lett., vol. 16, no. 11, pp. 973–976, Nov. 2009.
Comput. Vis., vol. 22, no. 3, pp. 199–218, Mar./Apr. 1997. [44] V. Duval, J. F. Aujol, and Y. Gousseau, “On the parameter choice for the
[38] J. A. Fessler, “Penalized weighted least-squares image reconstruction for non-local means,” Tech. Rep. HAL-00468856, 2010.
positron emission tomography,” IEEE Trans. Med. Imag., vol. 13, no. 2, [45] F. Xu and K. Mueller, “Accelerating popular tomographic reconstruction
pp. 290–300, Jun. 1994. algorithms on commodity PC graphics hardware,” IEEE Trans. Nucl. Sci.,
[39] W. P. Segars, G. Sturgeon, S. Mendonca, J. Grimes, and B. M. Tsui, “4D vol. 52, no. 3, pp. 654–663, Jun. 2005.
XCAT phantom for multimodality imaging research,” Med. Phys., vol. 37,
no. 9, pp. 4902–4915, Sep. 2010.
[40] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality
assessment: From error visibility to structural similarity,” IEEE Trans.
Image Process., vol. 13, no. 4, pp. 600–612, Apr. 2004. Authors’ photographs and biographies not available at the time of publication.