Abstract
Statistical image reconstruction in X-ray computed tomography yields large-scale regularized linear least-squares problems with nonnegativity bounds, where the memory footprint of the operator is a concern. Discretizing images in cylindrical coordinates results in significant memory savings, and allows parallel operator–vector products without on-the-fly computation of the operator, without necessarily decreasing image quality. However, it deteriorates the conditioning of the operator. We improve the Hessian conditioning by way of a block-circulant scaling operator and we propose a strategy to handle nondiagonal scaling in the context of projected-directions methods for bound-constrained problems. We describe our implementation of the scaling strategy using two algorithms: TRON, a trust-region method with exact second derivatives, and L-BFGS-B, a linesearch method with a limited-memory quasi-Newton Hessian approximation. We compare our approach with one where a change of variable is made in the problem. On two reconstruction problems, our approach converges faster than the change of variable approach, and achieves much tighter accuracy in terms of optimality residual than a first-order method.
Similar content being viewed by others
Notes
Available online at https://github.com/optimizers/NLPLab.
Available online at https://github.com/stephenbeckr/L-BFGS-B-C.
Available online at https://github.com/optimizers/NLPLab.
Available online at https://github.com/optimizers/NLPLab.
References
Ahn S, Fessler JA, Blatt D, Hero AO III (2006) Convergent incremental optimization transfer algorithms: application to tomography. IEEE Trans Med Imaging 25(3):283–296. https://doi.org/10.1109/TMI.2005.862740
Barzilai J, Borwein JM (1988) Two-point step size gradient methods. IMA J Numer Anal 8(1):141–148. https://doi.org/10.1093/imanum/8.1.141
Bertsekas DP (1982) Projected Newton methods for optimization problems with simple constraints. SIAM J Control Optim 20(2):221–246. https://doi.org/10.1137/0320018
Birgin EG, Martínez JM (2002) Large-scale active-set box-constrained optimization method with spectral projected gradients. Comput Optim Appl 23:101–125. https://doi.org/10.1023/A:1019928808826
Birgin EG, Martínez JM, Raydan M (2014) Spectral projected gradient methods: review and perspectives. J Stat Softw. https://doi.org/10.18637/jss.v060.i03
Bonettini S, Zanella R, Zanni L (2008) A scaled gradient projection method for constrained image deblurring. Inverse Probl 25(1):015002. https://doi.org/10.1088/0266-5611/25/1/015002
Bonettini S, Landi G, Piccolomini EL, Zanni L (2013) Scaling techniques for gradient projection-type methods in astronomical image deblurring. Int J Comput Math 90(1):9–29. https://doi.org/10.1080/00207160.2012.716513
Byrd RH, Nocedal J, Schnabel RB (1994) Representations of quasi-Newton matrices and their use in limited memory methods. Math Program 63(1–3):129–156. https://doi.org/10.1007/BF01582063
Byrd RH, Lu P, Nocedal J, Zhu C (1995) A limited memory algorithm for bound constrained optimization. SIAM J Sci Comput 16(5):1190–1208. https://doi.org/10.1137/0916069
Choi K, Wang J, Zhu L, Suh T, Boyd S, Xing L (2010) Compressed sensing based cone-beam computed tomography reconstruction with a first-order method. Math Program 37(9):5113–5125. https://doi.org/10.1118/1.3481510
Conn AR, Gould NIM, Toint PL (1988) Testing a class of methods for solving minimization problems with simple bounds on the variables. Math Comput 50(182):399–430. https://doi.org/10.2307/2008615 ISSN 0025-5718
Dolan ED, Moré JJ (2002) Benchmarking optimization software with performance profiles. Math Program A 91:201–213. https://doi.org/10.1007/s101070100263
Erdoǧan H, Fessler JA (1999a) Monotonic algorithms for transmission tomography. IEEE Trans Med Imaging 18(9):801–814. https://doi.org/10.1109/SSBI.2002.1233986
Erdoǧan H, Fessler JA (1999b) Ordered subsets algorithms for transmission tomography. Phys Med Biol 44(11):2835–2851. https://doi.org/10.1088/0031-9155/44/11/311
Feldkamp LA, Davis LC, Kress JW (1984) Practical cone-beam algorithm. J Opt Soc Am (A) 1(6):612–619. https://doi.org/10.1364/JOSAA.1.000612
Fessler JA (2000) Statistical image reconstruction methods for transmission tomography. In: Fritzpatrick JM, Sonka M (eds) Handbook of medical imaging, chapter 1, vol 2. SPIE Press, Bellingham, pp 1–70
Golkar M (2013) Fast iterative reconstruction in X-ray tomography using polar coordinates. PhD thesis, École Polytechnique de Montréal
Gould NI, Orban D, Toint PL (2015) CUTEst: a constrained and unconstrained testing environment with safe threads for mathematical optimization. Comput Optim Appl 60(3):545–557. https://doi.org/10.1007/s10589-014-9687-3
Goussard Y, Golkar M, Wagner A, Voorons M (2013) Cylindrical coordinate representation for statistical 3D CT reconstruction. In: Proceedings of international meeting on fully 3D image reconstruction in radiology and nuclear medicine, Lake Tahoe, CA, Jun, pp 138–141
Hamelin B (2009) Accélération d’une Approche Régularisée de Reconstruction en Tomographie à Rayons X avec Réduction des Artéfacts Métalliques. PhD thesis, École Polytechnique de Montréal
Hamelin B, Goussard Y, Dussault J-P, Cloutier G, Beaudoin G, Soulez G (2010) Design of iterative ROI transmission tomography reconstruction procedures and image quality analysis. Med Phys 37(9):4577–4589. https://doi.org/10.1118/1.3447722
Herman GT (2009) Fundamentals of computerized tomography: image reconstruction from projections. Springer, Berlin
Hestenes MR, Stiefel E (1952) Methods of conjugate gradients for solving linear systems. J Res Natl Bur Stand. https://doi.org/10.6028/jres.049.044
Hudson HM, Larkin RS (1994) Accelerated image reconstruction using ordered subsets of projection data. IEEE Trans Med Imaging 13(4):601–609. https://doi.org/10.1109/42.363108
Jensen TL, Jørgensen JH, Hansen PC, Jensen SH (2012) Implementation of an optimal first-order method for strongly convex total variation regularization. BIT Numer Math 52(2):329–356. https://doi.org/10.1007/s10543-011-0359-8 ISSN 1572-9125
Kim D, Ramani S, Fessler JA (2014) Combining ordered subsets and momentum for accelerated X-ray CT image reconstruction. IEEE Trans Med Imaging 34(1):167–178. https://doi.org/10.1109/TMI.2014.2350962
Kolehmainen V, Vanne A, Siltanen S, Jarvenpaa S, Kaipio JP, Lassas M, Kalke M (2006) Parallelized Bayesian inversion for three-dimensional dental X-ray imaging. IEEE Trans Med Imaging 25(2):218–228. https://doi.org/10.1109/TMI.2005.862662
Kolehmainen V, Vanne A, Siltanen S, Järvenpää S, Kaipio JP, Lassas M, Kalke M (2007) Bayesian inversion method for 3d dental X-ray imaging. e & i Elektrotechnik und Informationstechnik 124(7–8):248–253. https://doi.org/10.1007/s00502-007-0450-7
Landi G (2008) A projected Newton-CG method for nonnegative astronomical image deblurring. Numer Algorithms 48(4):279–300. https://doi.org/10.1007/s11075-008-9198-3 ISSN 1572-9265
Lange K, Fessler JA (1995) Globally convergent algorithms for maximum a posteriori transmission tomography. IEEE Trans Image Process 4(10):1430–1438. https://doi.org/10.1109/83.465107
Lin C-J, Moré JJ (1999) Newton’s method for large bound-constrained optimization problems. SIAM J Optim 9(4):1100–1127. https://doi.org/10.1137/S1052623498345075
McLaughlin M (2017) Méthodes sans factorisation pour la tomographie à rayons X en coordonnées cylindriques. Master’s thesis, École Polytechnique de Montréal, URL https://publications.polymtl.ca/2742
Morales JL, Nocedal J (2011) Remark on “Algorithm 778: L-BFGS-B: fortran subroutines for large-scale bound constrained optimization”. ACM Trans Math Softw (TOMS) 38(1):7. https://doi.org/10.1145/2049662.2049669
Moré JJ, Thuente DJ (1994) Line search algorithms with guaranteed sufficient decrease. ACM Trans Math Softw 20(3):286–307. https://doi.org/10.1145/192115.192132 ISSN 0098-3500
Nesterov YE (1983) A method for solving the convex programming problem with convergence rate \(O(1/k^2)\). Dokl Akad Nauk SSSR 269:543–547
Nien H, Fessler JA (2015) Fast X-ray CT image reconstruction using a linearized augmented Lagrangian method with ordered subsets. IEEE Trans Med Imaging 34(2):388–399. https://doi.org/10.1109/TMI.2014.2358499 ISSN 0278-0062
Nocedal J (1980) Updating quasi-Newton matrices with limited storage. Math Comput 35(151):773–782. https://doi.org/10.1090/S0025-5718-1980-0572855-7
Noo F, Hahn K, Schöndube H, Stierstorfer K (2016) Iterative CT reconstruction using coordinate descent with ordered subsets of data. In: Medical imaging 2016: physics of medical imaging, vol 9783. International Society for Optics and Photonics, p 97834A. https://doi.org/10.1117/12.2217558
Piccolomini EL, Coli V, Morotti E, Zanni L (2018) Reconstruction of 3D X-ray CT images from reduced sampling by a scaled gradient projection algorithm. Comput Optim Appl 71(1):171–191
Pock T, Chambolle A (2011) Diagonal preconditioning for first order primal-dual algorithms in convex optimization. In: International conference on computer vision. IEEE, pp 1762–1769. https://doi.org/10.1109/ICCV.2011.6126441
Sauer KD, Bouman CA (1993) A local update strategy for iterative reconstruction from projections. IEEE Trans Signal Process SP–41(2):534–548. https://doi.org/10.1109/78.193196
Segars WP, Mahesh M, Beck TJ, Frey EC, Tsui BMW (2008) Realistic CT simulation using the 4D XCAT phantom. Med Phys 35(8):3800–3808
Sidky EY, Jørgensen JS, Pan X (2013) First-order convex feasibility algorithms for X-ray CT. Med Phys. https://doi.org/10.1118/1.4790698
Thibaudeau C, Leroux J-D, Fontaine R, Lecomte R (2013) Fully 3D iterative CT reconstruction using polar coordinates. Med Phys 40(11):111904. https://doi.org/10.1118/1.4822514
Xu Q, Yang D, Tan J, Sawatzky A, Anastasio MA (2016) Accelerated fast iterative shrinkage thresholding algorithms for sparsity-regularized cone-beam CT image reconstruction. Med Phys 43(4):1849–1872. https://doi.org/10.1118/1.4942812
Zheng X, Ravishankar S, Long Y, Fessler JA (2018) PWLS-ULTRA: an efficient clustering and learning-based approach for low-dose 3D CT image reconstruction. IEEE Trans Med Imaging 37(6):1498–1510. https://doi.org/10.1109/TMI.2018.2832007
Zhu C, Byrd RH, Lu P, Nocedal J (1997) Algorithm 778. L-BFGS-B: fortran subroutines for large-scale bound constrained optimization. ACM Trans Math Softw 23(4):550–560. https://doi.org/10.1145/279232.279236
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Mestdagh, G., Goussard, Y. & Orban, D. Scaled projected-directions methods with application to transmission tomography. Optim Eng 21, 1537–1561 (2020). https://doi.org/10.1007/s11081-020-09484-0
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11081-020-09484-0