Markov-chain Monte Carlo for the performance of a channelized-ideal observer in detection tasks with non-Gaussian lumpy backgrounds

Subok Park, Eric W Clarkson

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

4 Citations (Scopus)

Abstract

The Bayesian ideal observer is optimal among all observers and sets an upper bound for observer performance in binary detection tasks.1 This observer provides a quantitative measure of diagnostic performance of an imaging system, summarized by the area under the receiver operating characteristic curve (AUC),1 and thus should be used for image quality assessment whenever possible. However, computation of ideal-observer performance is difficult because this observer requires the full description of the statistical properties of the signal-absent and signal-present data, which are often unknown in tasks involving complex backgrounds. Furthermore, the dimension of the integrals that need to be calculated for the observer is huge. To estimate ideal-observer performance in detection tasks with non-Gaussian lumpy backgrounds, Kupinski et al.2 developed a Markovchain Monte Carlo (MCMC) method, but this method has a disadvantage of long computation times. In an attempt to reduce the computation load and still approximate ideal-observer performance, Park et al.3,4 investigated a channelized-ideal observer (CIO) in similar tasks and found that the CIO with singular vectors of the imaging system approximated the performance of the ideal observer. But. in that work, an extension of the Kupinski MCMC was used for calculating the performance of the CIO and it did not reduce the computational burden. In the current work, we propose a new MCMC method, which we call a CIO-MCMC, to speed up the computation of the CIO. We use singular vectors of the imaging system as efficient channels for the ideal observer. Our results show that the CIO-MCMC has the potential to speed up the computation of ideal observer performance with a large number of channels.

Original languageEnglish (US)
Title of host publicationProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume6917
DOIs
StatePublished - 2008
EventMedical Imaging 2008 - Image Perception, Observer Performance, and Technology Assessment - San Diego, CA, United States
Duration: Feb 20 2008Feb 21 2008

Other

OtherMedical Imaging 2008 - Image Perception, Observer Performance, and Technology Assessment
CountryUnited States
CitySan Diego, CA
Period2/20/082/21/08

Fingerprint

Markov processes
Imaging systems
Monte Carlo methods
Image quality

Keywords

  • Bayesian ideal observer
  • Channelized-ideal observer
  • Efficient channels
  • Markov-chain Monte Carlo
  • Non-Gaussian lumpy backgrounds

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Markov-chain Monte Carlo for the performance of a channelized-ideal observer in detection tasks with non-Gaussian lumpy backgrounds. / Park, Subok; Clarkson, Eric W.

Progress in Biomedical Optics and Imaging - Proceedings of SPIE. Vol. 6917 2008. 69170T.

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

Park, S & Clarkson, EW 2008, Markov-chain Monte Carlo for the performance of a channelized-ideal observer in detection tasks with non-Gaussian lumpy backgrounds. in Progress in Biomedical Optics and Imaging - Proceedings of SPIE. vol. 6917, 69170T, Medical Imaging 2008 - Image Perception, Observer Performance, and Technology Assessment, San Diego, CA, United States, 2/20/08. https://doi.org/10.1117/12.771704
@inproceedings{83f7d791a84c4fab9c644f3529ce5c98,
title = "Markov-chain Monte Carlo for the performance of a channelized-ideal observer in detection tasks with non-Gaussian lumpy backgrounds",
abstract = "The Bayesian ideal observer is optimal among all observers and sets an upper bound for observer performance in binary detection tasks.1 This observer provides a quantitative measure of diagnostic performance of an imaging system, summarized by the area under the receiver operating characteristic curve (AUC),1 and thus should be used for image quality assessment whenever possible. However, computation of ideal-observer performance is difficult because this observer requires the full description of the statistical properties of the signal-absent and signal-present data, which are often unknown in tasks involving complex backgrounds. Furthermore, the dimension of the integrals that need to be calculated for the observer is huge. To estimate ideal-observer performance in detection tasks with non-Gaussian lumpy backgrounds, Kupinski et al.2 developed a Markovchain Monte Carlo (MCMC) method, but this method has a disadvantage of long computation times. In an attempt to reduce the computation load and still approximate ideal-observer performance, Park et al.3,4 investigated a channelized-ideal observer (CIO) in similar tasks and found that the CIO with singular vectors of the imaging system approximated the performance of the ideal observer. But. in that work, an extension of the Kupinski MCMC was used for calculating the performance of the CIO and it did not reduce the computational burden. In the current work, we propose a new MCMC method, which we call a CIO-MCMC, to speed up the computation of the CIO. We use singular vectors of the imaging system as efficient channels for the ideal observer. Our results show that the CIO-MCMC has the potential to speed up the computation of ideal observer performance with a large number of channels.",
keywords = "Bayesian ideal observer, Channelized-ideal observer, Efficient channels, Markov-chain Monte Carlo, Non-Gaussian lumpy backgrounds",
author = "Subok Park and Clarkson, {Eric W}",
year = "2008",
doi = "10.1117/12.771704",
language = "English (US)",
isbn = "9780819471017",
volume = "6917",
booktitle = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

}

TY - GEN

T1 - Markov-chain Monte Carlo for the performance of a channelized-ideal observer in detection tasks with non-Gaussian lumpy backgrounds

AU - Park, Subok

AU - Clarkson, Eric W

PY - 2008

Y1 - 2008

N2 - The Bayesian ideal observer is optimal among all observers and sets an upper bound for observer performance in binary detection tasks.1 This observer provides a quantitative measure of diagnostic performance of an imaging system, summarized by the area under the receiver operating characteristic curve (AUC),1 and thus should be used for image quality assessment whenever possible. However, computation of ideal-observer performance is difficult because this observer requires the full description of the statistical properties of the signal-absent and signal-present data, which are often unknown in tasks involving complex backgrounds. Furthermore, the dimension of the integrals that need to be calculated for the observer is huge. To estimate ideal-observer performance in detection tasks with non-Gaussian lumpy backgrounds, Kupinski et al.2 developed a Markovchain Monte Carlo (MCMC) method, but this method has a disadvantage of long computation times. In an attempt to reduce the computation load and still approximate ideal-observer performance, Park et al.3,4 investigated a channelized-ideal observer (CIO) in similar tasks and found that the CIO with singular vectors of the imaging system approximated the performance of the ideal observer. But. in that work, an extension of the Kupinski MCMC was used for calculating the performance of the CIO and it did not reduce the computational burden. In the current work, we propose a new MCMC method, which we call a CIO-MCMC, to speed up the computation of the CIO. We use singular vectors of the imaging system as efficient channels for the ideal observer. Our results show that the CIO-MCMC has the potential to speed up the computation of ideal observer performance with a large number of channels.

AB - The Bayesian ideal observer is optimal among all observers and sets an upper bound for observer performance in binary detection tasks.1 This observer provides a quantitative measure of diagnostic performance of an imaging system, summarized by the area under the receiver operating characteristic curve (AUC),1 and thus should be used for image quality assessment whenever possible. However, computation of ideal-observer performance is difficult because this observer requires the full description of the statistical properties of the signal-absent and signal-present data, which are often unknown in tasks involving complex backgrounds. Furthermore, the dimension of the integrals that need to be calculated for the observer is huge. To estimate ideal-observer performance in detection tasks with non-Gaussian lumpy backgrounds, Kupinski et al.2 developed a Markovchain Monte Carlo (MCMC) method, but this method has a disadvantage of long computation times. In an attempt to reduce the computation load and still approximate ideal-observer performance, Park et al.3,4 investigated a channelized-ideal observer (CIO) in similar tasks and found that the CIO with singular vectors of the imaging system approximated the performance of the ideal observer. But. in that work, an extension of the Kupinski MCMC was used for calculating the performance of the CIO and it did not reduce the computational burden. In the current work, we propose a new MCMC method, which we call a CIO-MCMC, to speed up the computation of the CIO. We use singular vectors of the imaging system as efficient channels for the ideal observer. Our results show that the CIO-MCMC has the potential to speed up the computation of ideal observer performance with a large number of channels.

KW - Bayesian ideal observer

KW - Channelized-ideal observer

KW - Efficient channels

KW - Markov-chain Monte Carlo

KW - Non-Gaussian lumpy backgrounds

UR - http://www.scopus.com/inward/record.url?scp=44949093599&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=44949093599&partnerID=8YFLogxK

U2 - 10.1117/12.771704

DO - 10.1117/12.771704

M3 - Conference contribution

SN - 9780819471017

VL - 6917

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

ER -