Magnetic resonance image analysis by information theoretic criteria and stochastic site models

Yue Wang, Tülay Adah, Jianhua Xuan, Zsolt Szabo

Research output: Contribution to journalArticle

Abstract

Quantitative analysis of magnetic resonance (MR) images is a powerful tool for image-guided diagnosis, monitoring, and intervention. The major tasks involve tissue quantification and image segmentation where both the pixel and context images are considered. To extract clinically useful information from images that might be lacking in prior knowledge, we introduce an unsupervised tissue characterization algorithm that is both statistically principled and patient specific. The method uses adaptive standard finite normal mixture and inhomogeneous Markov random field models, whose parameters are estimated using expectation-maximization and relaxation labeling algorithms under information theoretic criteria. We demonstrate the successful applications of the approach with synthetic data sets and then with real MR brain images.

Original languageEnglish (US)
Pages (from-to)150-158
Number of pages9
JournalIEEE Transactions on Information Technology in Biomedicine
Volume5
Issue number2
DOIs
StatePublished - Jun 2001

Fingerprint

Magnetic resonance
Image analysis
Magnetic Resonance Spectroscopy
Tissue
Image segmentation
Labeling
Brain
Pixels
Monitoring
Chemical analysis
Datasets

Keywords

  • Finite normal mixture
  • Image segmentation
  • Information theoretic criteria
  • Patient site model
  • Tissue quantification

ASJC Scopus subject areas

  • Health Informatics
  • Health Information Management
  • Information Systems
  • Computer Science Applications
  • Computational Theory and Mathematics

Cite this

Magnetic resonance image analysis by information theoretic criteria and stochastic site models. / Wang, Yue; Adah, Tülay; Xuan, Jianhua; Szabo, Zsolt.

In: IEEE Transactions on Information Technology in Biomedicine, Vol. 5, No. 2, 06.2001, p. 150-158.

Research output: Contribution to journalArticle

@article{92baaaf0a4d74c26a28178c5c8ac8ccd,
title = "Magnetic resonance image analysis by information theoretic criteria and stochastic site models",
abstract = "Quantitative analysis of magnetic resonance (MR) images is a powerful tool for image-guided diagnosis, monitoring, and intervention. The major tasks involve tissue quantification and image segmentation where both the pixel and context images are considered. To extract clinically useful information from images that might be lacking in prior knowledge, we introduce an unsupervised tissue characterization algorithm that is both statistically principled and patient specific. The method uses adaptive standard finite normal mixture and inhomogeneous Markov random field models, whose parameters are estimated using expectation-maximization and relaxation labeling algorithms under information theoretic criteria. We demonstrate the successful applications of the approach with synthetic data sets and then with real MR brain images.",
keywords = "Finite normal mixture, Image segmentation, Information theoretic criteria, Patient site model, Tissue quantification",
author = "Yue Wang and T{\"u}lay Adah and Jianhua Xuan and Zsolt Szabo",
year = "2001",
month = "6",
doi = "10.1109/4233.924805",
language = "English (US)",
volume = "5",
pages = "150--158",
journal = "IEEE Journal of Biomedical and Health Informatics",
issn = "2168-2194",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "2",

}

TY - JOUR

T1 - Magnetic resonance image analysis by information theoretic criteria and stochastic site models

AU - Wang, Yue

AU - Adah, Tülay

AU - Xuan, Jianhua

AU - Szabo, Zsolt

PY - 2001/6

Y1 - 2001/6

N2 - Quantitative analysis of magnetic resonance (MR) images is a powerful tool for image-guided diagnosis, monitoring, and intervention. The major tasks involve tissue quantification and image segmentation where both the pixel and context images are considered. To extract clinically useful information from images that might be lacking in prior knowledge, we introduce an unsupervised tissue characterization algorithm that is both statistically principled and patient specific. The method uses adaptive standard finite normal mixture and inhomogeneous Markov random field models, whose parameters are estimated using expectation-maximization and relaxation labeling algorithms under information theoretic criteria. We demonstrate the successful applications of the approach with synthetic data sets and then with real MR brain images.

AB - Quantitative analysis of magnetic resonance (MR) images is a powerful tool for image-guided diagnosis, monitoring, and intervention. The major tasks involve tissue quantification and image segmentation where both the pixel and context images are considered. To extract clinically useful information from images that might be lacking in prior knowledge, we introduce an unsupervised tissue characterization algorithm that is both statistically principled and patient specific. The method uses adaptive standard finite normal mixture and inhomogeneous Markov random field models, whose parameters are estimated using expectation-maximization and relaxation labeling algorithms under information theoretic criteria. We demonstrate the successful applications of the approach with synthetic data sets and then with real MR brain images.

KW - Finite normal mixture

KW - Image segmentation

KW - Information theoretic criteria

KW - Patient site model

KW - Tissue quantification

UR - http://www.scopus.com/inward/record.url?scp=0035354253&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035354253&partnerID=8YFLogxK

U2 - 10.1109/4233.924805

DO - 10.1109/4233.924805

M3 - Article

C2 - 11420993

AN - SCOPUS:0035354253

VL - 5

SP - 150

EP - 158

JO - IEEE Journal of Biomedical and Health Informatics

JF - IEEE Journal of Biomedical and Health Informatics

SN - 2168-2194

IS - 2

ER -