Faster PET reconstruction with a stochastic primal-dual hybrid gradient method

Matthias J. Ehrhardt, Pawel Markiewicz, Antonin Chambolle, Peter Richtárik, Jonathan Schott, Carola Bibiane Schönlieb

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

2 Citations (Scopus)

Abstract

Image reconstruction in positron emission tomography (PET) is computationally challenging due to Poisson noise, constraints and potentially non-smooth priors-let alone the sheer size of the problem. An algorithm that can cope well with the first three of the aforementioned challenges is the primal-dual hybrid gradient algorithm (PDHG) studied by Chambolle and Pock in 2011. However, PDHG updates all variables in parallel and is therefore computationally demanding on the large problem sizes encountered with modern PET scanners where the number of dual variables easily exceeds 100 million. In this work, we numerically study the usage of SPDHG-a stochastic extension of PDHG-but is still guaranteed to converge to a solution of the deterministic optimization problem with similar rates as PDHG. Numerical results on a clinical data set show that by introducing randomization into PDHG, similar results as the deterministic algorithm can be achieved using only around 10 % of operator evaluations. Thus, making significant progress towards the feasibility of sophisticated mathematical models in a clinical setting.

Original languageEnglish
Title of host publicationWavelets and Sparsity XVII
PublisherSPIE
ISBN (Electronic)9781510612457
DOIs
Publication statusPublished - 1 Jan 2017
EventWavelets and Sparsity XVII 2017 - San Diego, USA United States
Duration: 6 Aug 20179 Aug 2017

Publication series

NameProceedings of SPIE
Volume10394

Conference

ConferenceWavelets and Sparsity XVII 2017
CountryUSA United States
CitySan Diego
Period6/08/179/08/17

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Ehrhardt, M. J., Markiewicz, P., Chambolle, A., Richtárik, P., Schott, J., & Schönlieb, C. B. (2017). Faster PET reconstruction with a stochastic primal-dual hybrid gradient method. In Wavelets and Sparsity XVII [103941O] (Proceedings of SPIE; Vol. 10394). SPIE. https://doi.org/10.1117/12.2272946

Faster PET reconstruction with a stochastic primal-dual hybrid gradient method. / Ehrhardt, Matthias J.; Markiewicz, Pawel; Chambolle, Antonin; Richtárik, Peter; Schott, Jonathan; Schönlieb, Carola Bibiane.

Wavelets and Sparsity XVII. SPIE, 2017. 103941O (Proceedings of SPIE; Vol. 10394).

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

Ehrhardt, MJ, Markiewicz, P, Chambolle, A, Richtárik, P, Schott, J & Schönlieb, CB 2017, Faster PET reconstruction with a stochastic primal-dual hybrid gradient method. in Wavelets and Sparsity XVII., 103941O, Proceedings of SPIE, vol. 10394, SPIE, Wavelets and Sparsity XVII 2017, San Diego, USA United States, 6/08/17. https://doi.org/10.1117/12.2272946
Ehrhardt MJ, Markiewicz P, Chambolle A, Richtárik P, Schott J, Schönlieb CB. Faster PET reconstruction with a stochastic primal-dual hybrid gradient method. In Wavelets and Sparsity XVII. SPIE. 2017. 103941O. (Proceedings of SPIE). https://doi.org/10.1117/12.2272946
Ehrhardt, Matthias J. ; Markiewicz, Pawel ; Chambolle, Antonin ; Richtárik, Peter ; Schott, Jonathan ; Schönlieb, Carola Bibiane. / Faster PET reconstruction with a stochastic primal-dual hybrid gradient method. Wavelets and Sparsity XVII. SPIE, 2017. (Proceedings of SPIE).
@inproceedings{834652886b824848a9e5e264df5f6823,
title = "Faster PET reconstruction with a stochastic primal-dual hybrid gradient method",
abstract = "Image reconstruction in positron emission tomography (PET) is computationally challenging due to Poisson noise, constraints and potentially non-smooth priors-let alone the sheer size of the problem. An algorithm that can cope well with the first three of the aforementioned challenges is the primal-dual hybrid gradient algorithm (PDHG) studied by Chambolle and Pock in 2011. However, PDHG updates all variables in parallel and is therefore computationally demanding on the large problem sizes encountered with modern PET scanners where the number of dual variables easily exceeds 100 million. In this work, we numerically study the usage of SPDHG-a stochastic extension of PDHG-but is still guaranteed to converge to a solution of the deterministic optimization problem with similar rates as PDHG. Numerical results on a clinical data set show that by introducing randomization into PDHG, similar results as the deterministic algorithm can be achieved using only around 10 {\%} of operator evaluations. Thus, making significant progress towards the feasibility of sophisticated mathematical models in a clinical setting.",
author = "Ehrhardt, {Matthias J.} and Pawel Markiewicz and Antonin Chambolle and Peter Richt{\'a}rik and Jonathan Schott and Sch{\"o}nlieb, {Carola Bibiane}",
year = "2017",
month = "1",
day = "1",
doi = "10.1117/12.2272946",
language = "English",
series = "Proceedings of SPIE",
publisher = "SPIE",
booktitle = "Wavelets and Sparsity XVII",
address = "USA United States",

}

TY - GEN

T1 - Faster PET reconstruction with a stochastic primal-dual hybrid gradient method

AU - Ehrhardt, Matthias J.

AU - Markiewicz, Pawel

AU - Chambolle, Antonin

AU - Richtárik, Peter

AU - Schott, Jonathan

AU - Schönlieb, Carola Bibiane

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Image reconstruction in positron emission tomography (PET) is computationally challenging due to Poisson noise, constraints and potentially non-smooth priors-let alone the sheer size of the problem. An algorithm that can cope well with the first three of the aforementioned challenges is the primal-dual hybrid gradient algorithm (PDHG) studied by Chambolle and Pock in 2011. However, PDHG updates all variables in parallel and is therefore computationally demanding on the large problem sizes encountered with modern PET scanners where the number of dual variables easily exceeds 100 million. In this work, we numerically study the usage of SPDHG-a stochastic extension of PDHG-but is still guaranteed to converge to a solution of the deterministic optimization problem with similar rates as PDHG. Numerical results on a clinical data set show that by introducing randomization into PDHG, similar results as the deterministic algorithm can be achieved using only around 10 % of operator evaluations. Thus, making significant progress towards the feasibility of sophisticated mathematical models in a clinical setting.

AB - Image reconstruction in positron emission tomography (PET) is computationally challenging due to Poisson noise, constraints and potentially non-smooth priors-let alone the sheer size of the problem. An algorithm that can cope well with the first three of the aforementioned challenges is the primal-dual hybrid gradient algorithm (PDHG) studied by Chambolle and Pock in 2011. However, PDHG updates all variables in parallel and is therefore computationally demanding on the large problem sizes encountered with modern PET scanners where the number of dual variables easily exceeds 100 million. In this work, we numerically study the usage of SPDHG-a stochastic extension of PDHG-but is still guaranteed to converge to a solution of the deterministic optimization problem with similar rates as PDHG. Numerical results on a clinical data set show that by introducing randomization into PDHG, similar results as the deterministic algorithm can be achieved using only around 10 % of operator evaluations. Thus, making significant progress towards the feasibility of sophisticated mathematical models in a clinical setting.

UR - http://www.scopus.com/inward/record.url?scp=85033555990&partnerID=8YFLogxK

U2 - 10.1117/12.2272946

DO - 10.1117/12.2272946

M3 - Conference contribution

T3 - Proceedings of SPIE

BT - Wavelets and Sparsity XVII

PB - SPIE

ER -