A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This work studies the dual formulation of a penalized maximum likelihood reconstruction problem in x-ray CT. The primal objective function is a Poisson log-likelihood combined with a weighted cross-entropy penalty term. The dual formulation of the primal optimization problem is then derived and the optimization procedure outlined. The dual formulation better exploits the structure of the problem, which translates to faster convergence of iterative reconstruction algorithms. A gradient descent algorithm is implemented for solving the dual problem and its performance is compared with the filtered back-projection algorithm, and with the primal formulation optimized by using surrogate functions. The 3D XCAT phantom and an analytical x-ray CT simulator are used to generate noise-free and noisy CT projection data set with monochromatic and polychromatic x-ray spectrums. The reconstructed images from the dual formulation delineate the internal structures at early iterations better than the primal formulation using surrogate functions. However the body contour is slower to converge in the dual than in the primal formulation. The dual formulation demonstrate better noise-resolution tradeoff near the internal organs than the primal formulation. Since the surrogate functions in general can provide a diagonal approximation of the Hessian matrix of the objective function, further convergence speed up may be achieved by deriving the surrogate function of the dual objective function.

Original languageEnglish (US)
Title of host publicationProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume7258
DOIs
StatePublished - 2009
EventMedical Imaging 2009: Physics of Medical Imaging - Lake Buena Vista, FL, United States
Duration: Feb 9 2009Feb 12 2009

Other

OtherMedical Imaging 2009: Physics of Medical Imaging
CountryUnited States
CityLake Buena Vista, FL
Period2/9/092/12/09

Fingerprint

Maximum likelihood
X-Rays
formulations
X rays
Noise
x rays
Entropy
projection
Hessian matrices
optimization
x ray spectra
descent
tradeoffs
penalties
Simulators
organs
simulators
iteration
entropy
gradients

Keywords

  • Computed tomography
  • Dual optimization
  • Emission computed tomography
  • Poisson likelihood
  • Primal optimization
  • Rreconstruction algorithm
  • X-ray CT

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Cite this

Xu, J., Taguchi, K., Gullberg, G. T., & Tsui, B. (2009). A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem. In Progress in Biomedical Optics and Imaging - Proceedings of SPIE (Vol. 7258). [725828] https://doi.org/10.1117/12.813873

A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem. / Xu, Jingyan; Taguchi, Katsuyuki; Gullberg, Grant T.; Tsui, Benjamin.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 7258 2009. 725828.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Xu, J, Taguchi, K, Gullberg, GT & Tsui, B 2009, A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem. in Progress in Biomedical Optics and Imaging - Proceedings of SPIE. vol. 7258, 725828, Medical Imaging 2009: Physics of Medical Imaging, Lake Buena Vista, FL, United States, 2/9/09. https://doi.org/10.1117/12.813873
Xu J, Taguchi K, Gullberg GT, Tsui B. A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem. In Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 7258. 2009. 725828 https://doi.org/10.1117/12.813873
Xu, Jingyan ; Taguchi, Katsuyuki ; Gullberg, Grant T. ; Tsui, Benjamin. / A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem. Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 7258 2009.
@inproceedings{a88d14c0163d4dea96d21149d8be1f61,
title = "A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem",
abstract = "This work studies the dual formulation of a penalized maximum likelihood reconstruction problem in x-ray CT. The primal objective function is a Poisson log-likelihood combined with a weighted cross-entropy penalty term. The dual formulation of the primal optimization problem is then derived and the optimization procedure outlined. The dual formulation better exploits the structure of the problem, which translates to faster convergence of iterative reconstruction algorithms. A gradient descent algorithm is implemented for solving the dual problem and its performance is compared with the filtered back-projection algorithm, and with the primal formulation optimized by using surrogate functions. The 3D XCAT phantom and an analytical x-ray CT simulator are used to generate noise-free and noisy CT projection data set with monochromatic and polychromatic x-ray spectrums. The reconstructed images from the dual formulation delineate the internal structures at early iterations better than the primal formulation using surrogate functions. However the body contour is slower to converge in the dual than in the primal formulation. The dual formulation demonstrate better noise-resolution tradeoff near the internal organs than the primal formulation. Since the surrogate functions in general can provide a diagonal approximation of the Hessian matrix of the objective function, further convergence speed up may be achieved by deriving the surrogate function of the dual objective function.",
keywords = "Computed tomography, Dual optimization, Emission computed tomography, Poisson likelihood, Primal optimization, Rreconstruction algorithm, X-ray CT",
author = "Jingyan Xu and Katsuyuki Taguchi and Gullberg, {Grant T.} and Benjamin Tsui",
year = "2009",
doi = "10.1117/12.813873",
language = "English (US)",
isbn = "9780819475091",
volume = "7258",
booktitle = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

}

TY - GEN

T1 - A dual formulation of a penalized maximum likelihood x-ray CT reconstruction Problem

AU - Xu, Jingyan

AU - Taguchi, Katsuyuki

AU - Gullberg, Grant T.

AU - Tsui, Benjamin

PY - 2009

Y1 - 2009

N2 - This work studies the dual formulation of a penalized maximum likelihood reconstruction problem in x-ray CT. The primal objective function is a Poisson log-likelihood combined with a weighted cross-entropy penalty term. The dual formulation of the primal optimization problem is then derived and the optimization procedure outlined. The dual formulation better exploits the structure of the problem, which translates to faster convergence of iterative reconstruction algorithms. A gradient descent algorithm is implemented for solving the dual problem and its performance is compared with the filtered back-projection algorithm, and with the primal formulation optimized by using surrogate functions. The 3D XCAT phantom and an analytical x-ray CT simulator are used to generate noise-free and noisy CT projection data set with monochromatic and polychromatic x-ray spectrums. The reconstructed images from the dual formulation delineate the internal structures at early iterations better than the primal formulation using surrogate functions. However the body contour is slower to converge in the dual than in the primal formulation. The dual formulation demonstrate better noise-resolution tradeoff near the internal organs than the primal formulation. Since the surrogate functions in general can provide a diagonal approximation of the Hessian matrix of the objective function, further convergence speed up may be achieved by deriving the surrogate function of the dual objective function.

AB - This work studies the dual formulation of a penalized maximum likelihood reconstruction problem in x-ray CT. The primal objective function is a Poisson log-likelihood combined with a weighted cross-entropy penalty term. The dual formulation of the primal optimization problem is then derived and the optimization procedure outlined. The dual formulation better exploits the structure of the problem, which translates to faster convergence of iterative reconstruction algorithms. A gradient descent algorithm is implemented for solving the dual problem and its performance is compared with the filtered back-projection algorithm, and with the primal formulation optimized by using surrogate functions. The 3D XCAT phantom and an analytical x-ray CT simulator are used to generate noise-free and noisy CT projection data set with monochromatic and polychromatic x-ray spectrums. The reconstructed images from the dual formulation delineate the internal structures at early iterations better than the primal formulation using surrogate functions. However the body contour is slower to converge in the dual than in the primal formulation. The dual formulation demonstrate better noise-resolution tradeoff near the internal organs than the primal formulation. Since the surrogate functions in general can provide a diagonal approximation of the Hessian matrix of the objective function, further convergence speed up may be achieved by deriving the surrogate function of the dual objective function.

KW - Computed tomography

KW - Dual optimization

KW - Emission computed tomography

KW - Poisson likelihood

KW - Primal optimization

KW - Rreconstruction algorithm

KW - X-ray CT

UR - http://www.scopus.com/inward/record.url?scp=66749106364&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=66749106364&partnerID=8YFLogxK

U2 - 10.1117/12.813873

DO - 10.1117/12.813873

M3 - Conference contribution

AN - SCOPUS:66749106364

SN - 9780819475091

VL - 7258

BT - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

ER -