Evaluation of image registration spatial accuracy using a Bayesian hierarchical model

Suyu Liu, Ying Yuan, Richard Castillo, Thomas Guerrero, Valen E. Johnson

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

To evaluate the utility of automated deformable image registration (DIR) algorithms, it is necessary to evaluate both the registration accuracy of the DIR algorithm itself, as well as the registration accuracy of the human readers from whom the "gold standard" is obtained. We propose a Bayesian hierarchical model to evaluate the spatial accuracy of human readers and automatic DIR methods based on multiple image registration data generated by human readers and automatic DIR methods. To fully account for the locations of landmarks in all images, we treat the true locations of landmarks as latent variables and impose a hierarchical structure on the magnitude of registration errors observed across image pairs. DIR registration errors are modeled using Gaussian processes with reference prior densities on prior parameters that determine the associated covariance matrices. We develop a Gibbs sampling algorithm to efficiently fit our models to high-dimensional data, and apply the proposed method to analyze an image dataset obtained from a 4D thoracic CT study.

Original languageEnglish (US)
Pages (from-to)366-377
Number of pages12
JournalBiometrics
Volume70
Issue number2
DOIs
StatePublished - 2014
Externally publishedYes

Fingerprint

Bayesian Hierarchical Model
Image registration
Image Registration
Registration
Evaluation
Four-Dimensional Computed Tomography
chest
Landmarks
gold
Evaluate
Thorax
methodology
Reference Prior
Gibbs Sampling
Latent Variables
High-dimensional Data
Hierarchical Structure
Covariance matrix
Gaussian Process
Gold

Keywords

  • Bayesian analysis
  • Image processing
  • Latent variable
  • Spatial correlation

ASJC Scopus subject areas

  • Applied Mathematics
  • Statistics and Probability
  • Agricultural and Biological Sciences(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Immunology and Microbiology(all)
  • Medicine(all)

Cite this

Liu, S., Yuan, Y., Castillo, R., Guerrero, T., & Johnson, V. E. (2014). Evaluation of image registration spatial accuracy using a Bayesian hierarchical model. Biometrics, 70(2), 366-377. https://doi.org/10.1111/biom.12146

Evaluation of image registration spatial accuracy using a Bayesian hierarchical model. / Liu, Suyu; Yuan, Ying; Castillo, Richard; Guerrero, Thomas; Johnson, Valen E.

In: Biometrics, Vol. 70, No. 2, 2014, p. 366-377.

Research output: Contribution to journalArticle

Liu, S, Yuan, Y, Castillo, R, Guerrero, T & Johnson, VE 2014, 'Evaluation of image registration spatial accuracy using a Bayesian hierarchical model', Biometrics, vol. 70, no. 2, pp. 366-377. https://doi.org/10.1111/biom.12146
Liu, Suyu ; Yuan, Ying ; Castillo, Richard ; Guerrero, Thomas ; Johnson, Valen E. / Evaluation of image registration spatial accuracy using a Bayesian hierarchical model. In: Biometrics. 2014 ; Vol. 70, No. 2. pp. 366-377.
@article{79125a55d02f43d2b57acfe3ea3cacf5,
title = "Evaluation of image registration spatial accuracy using a Bayesian hierarchical model",
abstract = "To evaluate the utility of automated deformable image registration (DIR) algorithms, it is necessary to evaluate both the registration accuracy of the DIR algorithm itself, as well as the registration accuracy of the human readers from whom the {"}gold standard{"} is obtained. We propose a Bayesian hierarchical model to evaluate the spatial accuracy of human readers and automatic DIR methods based on multiple image registration data generated by human readers and automatic DIR methods. To fully account for the locations of landmarks in all images, we treat the true locations of landmarks as latent variables and impose a hierarchical structure on the magnitude of registration errors observed across image pairs. DIR registration errors are modeled using Gaussian processes with reference prior densities on prior parameters that determine the associated covariance matrices. We develop a Gibbs sampling algorithm to efficiently fit our models to high-dimensional data, and apply the proposed method to analyze an image dataset obtained from a 4D thoracic CT study.",
keywords = "Bayesian analysis, Image processing, Latent variable, Spatial correlation",
author = "Suyu Liu and Ying Yuan and Richard Castillo and Thomas Guerrero and Johnson, {Valen E.}",
year = "2014",
doi = "10.1111/biom.12146",
language = "English (US)",
volume = "70",
pages = "366--377",
journal = "Biometrics",
issn = "0006-341X",
publisher = "Wiley-Blackwell",
number = "2",

}

TY - JOUR

T1 - Evaluation of image registration spatial accuracy using a Bayesian hierarchical model

AU - Liu, Suyu

AU - Yuan, Ying

AU - Castillo, Richard

AU - Guerrero, Thomas

AU - Johnson, Valen E.

PY - 2014

Y1 - 2014

N2 - To evaluate the utility of automated deformable image registration (DIR) algorithms, it is necessary to evaluate both the registration accuracy of the DIR algorithm itself, as well as the registration accuracy of the human readers from whom the "gold standard" is obtained. We propose a Bayesian hierarchical model to evaluate the spatial accuracy of human readers and automatic DIR methods based on multiple image registration data generated by human readers and automatic DIR methods. To fully account for the locations of landmarks in all images, we treat the true locations of landmarks as latent variables and impose a hierarchical structure on the magnitude of registration errors observed across image pairs. DIR registration errors are modeled using Gaussian processes with reference prior densities on prior parameters that determine the associated covariance matrices. We develop a Gibbs sampling algorithm to efficiently fit our models to high-dimensional data, and apply the proposed method to analyze an image dataset obtained from a 4D thoracic CT study.

AB - To evaluate the utility of automated deformable image registration (DIR) algorithms, it is necessary to evaluate both the registration accuracy of the DIR algorithm itself, as well as the registration accuracy of the human readers from whom the "gold standard" is obtained. We propose a Bayesian hierarchical model to evaluate the spatial accuracy of human readers and automatic DIR methods based on multiple image registration data generated by human readers and automatic DIR methods. To fully account for the locations of landmarks in all images, we treat the true locations of landmarks as latent variables and impose a hierarchical structure on the magnitude of registration errors observed across image pairs. DIR registration errors are modeled using Gaussian processes with reference prior densities on prior parameters that determine the associated covariance matrices. We develop a Gibbs sampling algorithm to efficiently fit our models to high-dimensional data, and apply the proposed method to analyze an image dataset obtained from a 4D thoracic CT study.

KW - Bayesian analysis

KW - Image processing

KW - Latent variable

KW - Spatial correlation

UR - http://www.scopus.com/inward/record.url?scp=84902332554&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84902332554&partnerID=8YFLogxK

U2 - 10.1111/biom.12146

DO - 10.1111/biom.12146

M3 - Article

C2 - 24575781

AN - SCOPUS:84902332554

VL - 70

SP - 366

EP - 377

JO - Biometrics

JF - Biometrics

SN - 0006-341X

IS - 2

ER -