Detection of prostate cancer with multiparametric MRI utilizing the anatomic structure of the prostate

Jin Jin, Lin Zhang, Ethan Leng, Greg Metzger, Joe Koopmeiners

Research output: Contribution to journalArticle

Abstract

Multiparametric magnetic resonance imaging (mpMRI), which combines traditional anatomic and newer quantitative MRI methods, has been shown to result in improved voxel-wise classification of prostate cancer as compared with any single MRI parameter. While these results are promising, substantial heterogeneity in the mpMRI parameter values and voxel-wise prostate cancer risk has been observed both between and within regions of the prostate. This suggests that classification of prostate cancer can potentially be improved by incorporating structural information into the classifier. In this paper, we propose a novel voxel-wise classifier of prostate cancer that accounts for the anatomic structure of the prostate by Bayesian hierarchical modeling, which can be combined with post hoc spatial Gaussian kernel smoothing to account for residual spatial correlation. Our proposed classifier results in significantly improved area under the ROC curve (0.822 vs 0.729, P <.001) and sensitivity corresponding to 90% specificity (0.599 vs 0.429, P <.001), compared with a baseline model that does not account for the anatomic structure of the prostate. Furthermore, the classifier can also be applied on voxels with missing mpMRI parameters, resulting in similar performance, which is an important practical consideration that cannot be easily accommodated using regression-based classifiers. In addition, our classifier achieved high computational efficiency with a closed-form solution for the posterior predictive cancer probability.

Original languageEnglish (US)
Pages (from-to)3214-3229
Number of pages16
JournalStatistics in Medicine
Volume37
Issue number22
DOIs
StatePublished - Sep 30 2018

Fingerprint

Prostate Cancer
Prostate
Prostatic Neoplasms
Classifier
Voxel
Magnetic Resonance Imaging
ROC Curve
Area Under Curve
Kernel Smoothing
Hierarchical Modeling
Bayesian Modeling
Gaussian Kernel
Receiver Operating Characteristic Curve
Spatial Correlation
Closed-form Solution
Computational Efficiency
Specificity
High Efficiency
Baseline
Cancer

Keywords

  • Bayesian classifier
  • multiparametric magnetic resonance imaging
  • prostate cancer
  • spatial classifier
  • voxel-wise classification

PubMed: MeSH publication types

  • Journal Article
  • Research Support, U.S. Gov't, Non-P.H.S.
  • Research Support, N.I.H., Extramural

Cite this

Detection of prostate cancer with multiparametric MRI utilizing the anatomic structure of the prostate. / Jin, Jin; Zhang, Lin; Leng, Ethan; Metzger, Greg; Koopmeiners, Joe.

In: Statistics in Medicine, Vol. 37, No. 22, 30.09.2018, p. 3214-3229.

Research output: Contribution to journalArticle

@article{90072449c17142379740a68c08365675,
title = "Detection of prostate cancer with multiparametric MRI utilizing the anatomic structure of the prostate",
abstract = "Multiparametric magnetic resonance imaging (mpMRI), which combines traditional anatomic and newer quantitative MRI methods, has been shown to result in improved voxel-wise classification of prostate cancer as compared with any single MRI parameter. While these results are promising, substantial heterogeneity in the mpMRI parameter values and voxel-wise prostate cancer risk has been observed both between and within regions of the prostate. This suggests that classification of prostate cancer can potentially be improved by incorporating structural information into the classifier. In this paper, we propose a novel voxel-wise classifier of prostate cancer that accounts for the anatomic structure of the prostate by Bayesian hierarchical modeling, which can be combined with post hoc spatial Gaussian kernel smoothing to account for residual spatial correlation. Our proposed classifier results in significantly improved area under the ROC curve (0.822 vs 0.729, P <.001) and sensitivity corresponding to 90{\%} specificity (0.599 vs 0.429, P <.001), compared with a baseline model that does not account for the anatomic structure of the prostate. Furthermore, the classifier can also be applied on voxels with missing mpMRI parameters, resulting in similar performance, which is an important practical consideration that cannot be easily accommodated using regression-based classifiers. In addition, our classifier achieved high computational efficiency with a closed-form solution for the posterior predictive cancer probability.",
keywords = "Bayesian classifier, multiparametric magnetic resonance imaging, prostate cancer, spatial classifier, voxel-wise classification",
author = "Jin Jin and Lin Zhang and Ethan Leng and Greg Metzger and Joe Koopmeiners",
year = "2018",
month = "9",
day = "30",
doi = "10.1002/sim.7810",
language = "English (US)",
volume = "37",
pages = "3214--3229",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",
number = "22",

}

TY - JOUR

T1 - Detection of prostate cancer with multiparametric MRI utilizing the anatomic structure of the prostate

AU - Jin, Jin

AU - Zhang, Lin

AU - Leng, Ethan

AU - Metzger, Greg

AU - Koopmeiners, Joe

PY - 2018/9/30

Y1 - 2018/9/30

N2 - Multiparametric magnetic resonance imaging (mpMRI), which combines traditional anatomic and newer quantitative MRI methods, has been shown to result in improved voxel-wise classification of prostate cancer as compared with any single MRI parameter. While these results are promising, substantial heterogeneity in the mpMRI parameter values and voxel-wise prostate cancer risk has been observed both between and within regions of the prostate. This suggests that classification of prostate cancer can potentially be improved by incorporating structural information into the classifier. In this paper, we propose a novel voxel-wise classifier of prostate cancer that accounts for the anatomic structure of the prostate by Bayesian hierarchical modeling, which can be combined with post hoc spatial Gaussian kernel smoothing to account for residual spatial correlation. Our proposed classifier results in significantly improved area under the ROC curve (0.822 vs 0.729, P <.001) and sensitivity corresponding to 90% specificity (0.599 vs 0.429, P <.001), compared with a baseline model that does not account for the anatomic structure of the prostate. Furthermore, the classifier can also be applied on voxels with missing mpMRI parameters, resulting in similar performance, which is an important practical consideration that cannot be easily accommodated using regression-based classifiers. In addition, our classifier achieved high computational efficiency with a closed-form solution for the posterior predictive cancer probability.

AB - Multiparametric magnetic resonance imaging (mpMRI), which combines traditional anatomic and newer quantitative MRI methods, has been shown to result in improved voxel-wise classification of prostate cancer as compared with any single MRI parameter. While these results are promising, substantial heterogeneity in the mpMRI parameter values and voxel-wise prostate cancer risk has been observed both between and within regions of the prostate. This suggests that classification of prostate cancer can potentially be improved by incorporating structural information into the classifier. In this paper, we propose a novel voxel-wise classifier of prostate cancer that accounts for the anatomic structure of the prostate by Bayesian hierarchical modeling, which can be combined with post hoc spatial Gaussian kernel smoothing to account for residual spatial correlation. Our proposed classifier results in significantly improved area under the ROC curve (0.822 vs 0.729, P <.001) and sensitivity corresponding to 90% specificity (0.599 vs 0.429, P <.001), compared with a baseline model that does not account for the anatomic structure of the prostate. Furthermore, the classifier can also be applied on voxels with missing mpMRI parameters, resulting in similar performance, which is an important practical consideration that cannot be easily accommodated using regression-based classifiers. In addition, our classifier achieved high computational efficiency with a closed-form solution for the posterior predictive cancer probability.

KW - Bayesian classifier

KW - multiparametric magnetic resonance imaging

KW - prostate cancer

KW - spatial classifier

KW - voxel-wise classification

UR - http://www.scopus.com/inward/record.url?scp=85052730584&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85052730584&partnerID=8YFLogxK

U2 - 10.1002/sim.7810

DO - 10.1002/sim.7810

M3 - Article

VL - 37

SP - 3214

EP - 3229

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 22

ER -