MR image segmentation using a power transformation approach

Juin Der Lee, Hong Ren Su, Philip E. Cheng, Michelle Liou, John A D Aston, Arthur C. Tsai, Cheng Yu Chen

Research output: Contribution to journalArticle

48 Citations (Scopus)

Abstract

This study proposes a segmentation method for brain MR images using a distribution transformation approach. The method extends traditional Gaussian mixtures expectation-maximization segmentation to a power transformed version of mixed intensity distributions, which includes Gaussian mixtures as a special case. As MR intensities tend to exhibit non-Gaussianity due to partial volume effects, the proposed method is designed to fit non-Gaussian tissue intensity distributions. One advantage of the method is that it is intuitively appealing and computationally simple. To avoid performance degradation caused by intensity inhomogeneity, different methods for correcting bias fields were applied prior to image segmentation, and their correction effects on the segmentation results were examined in the empirical study. The partitions of brain tissues (i.e., gray and white matter) resulting from the method were validated and evaluated against manual segmentation results based on 38 real T1-weighted image volumes from the internet brain segmentation repository, and 18 simulated image volumes from Brain Web. The Jaccard and Dice similarity indexes were computed to evaluate the performance of the proposed approach relative to the expert segmentations. Empirical results suggested that the proposed segmentation method yielded higher similarity measures for both gray matter and white matter as compared with those based on the traditional segmentation using the Gaussian mixtures approach.

Original languageEnglish
Article number4752750
Pages (from-to)894-905
Number of pages12
JournalIEEE Transactions on Medical Imaging
Volume28
Issue number6
DOIs
Publication statusPublished - Jun 2009
Externally publishedYes

Fingerprint

Image segmentation
Brain
Tissue
Internet
Degradation
Tissue Distribution

Keywords

  • Box-cox transformation
  • Expectation-maximization (EM) algorithm
  • Gaussian mixtures
  • Statistical segmentation

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Computer Science Applications
  • Radiological and Ultrasound Technology
  • Software

Cite this

Lee, J. D., Su, H. R., Cheng, P. E., Liou, M., Aston, J. A. D., Tsai, A. C., & Chen, C. Y. (2009). MR image segmentation using a power transformation approach. IEEE Transactions on Medical Imaging, 28(6), 894-905. [4752750]. https://doi.org/10.1109/TMI.2009.2012896

MR image segmentation using a power transformation approach. / Lee, Juin Der; Su, Hong Ren; Cheng, Philip E.; Liou, Michelle; Aston, John A D; Tsai, Arthur C.; Chen, Cheng Yu.

In: IEEE Transactions on Medical Imaging, Vol. 28, No. 6, 4752750, 06.2009, p. 894-905.

Research output: Contribution to journalArticle

Lee, JD, Su, HR, Cheng, PE, Liou, M, Aston, JAD, Tsai, AC & Chen, CY 2009, 'MR image segmentation using a power transformation approach', IEEE Transactions on Medical Imaging, vol. 28, no. 6, 4752750, pp. 894-905. https://doi.org/10.1109/TMI.2009.2012896
Lee JD, Su HR, Cheng PE, Liou M, Aston JAD, Tsai AC et al. MR image segmentation using a power transformation approach. IEEE Transactions on Medical Imaging. 2009 Jun;28(6):894-905. 4752750. https://doi.org/10.1109/TMI.2009.2012896
Lee, Juin Der ; Su, Hong Ren ; Cheng, Philip E. ; Liou, Michelle ; Aston, John A D ; Tsai, Arthur C. ; Chen, Cheng Yu. / MR image segmentation using a power transformation approach. In: IEEE Transactions on Medical Imaging. 2009 ; Vol. 28, No. 6. pp. 894-905.
@article{a065231ef93b40219b35289a4e264f21,
title = "MR image segmentation using a power transformation approach",
abstract = "This study proposes a segmentation method for brain MR images using a distribution transformation approach. The method extends traditional Gaussian mixtures expectation-maximization segmentation to a power transformed version of mixed intensity distributions, which includes Gaussian mixtures as a special case. As MR intensities tend to exhibit non-Gaussianity due to partial volume effects, the proposed method is designed to fit non-Gaussian tissue intensity distributions. One advantage of the method is that it is intuitively appealing and computationally simple. To avoid performance degradation caused by intensity inhomogeneity, different methods for correcting bias fields were applied prior to image segmentation, and their correction effects on the segmentation results were examined in the empirical study. The partitions of brain tissues (i.e., gray and white matter) resulting from the method were validated and evaluated against manual segmentation results based on 38 real T1-weighted image volumes from the internet brain segmentation repository, and 18 simulated image volumes from Brain Web. The Jaccard and Dice similarity indexes were computed to evaluate the performance of the proposed approach relative to the expert segmentations. Empirical results suggested that the proposed segmentation method yielded higher similarity measures for both gray matter and white matter as compared with those based on the traditional segmentation using the Gaussian mixtures approach.",
keywords = "Box-cox transformation, Expectation-maximization (EM) algorithm, Gaussian mixtures, Statistical segmentation",
author = "Lee, {Juin Der} and Su, {Hong Ren} and Cheng, {Philip E.} and Michelle Liou and Aston, {John A D} and Tsai, {Arthur C.} and Chen, {Cheng Yu}",
year = "2009",
month = "6",
doi = "10.1109/TMI.2009.2012896",
language = "English",
volume = "28",
pages = "894--905",
journal = "IEEE Transactions on Medical Imaging",
issn = "0278-0062",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "6",

}

TY - JOUR

T1 - MR image segmentation using a power transformation approach

AU - Lee, Juin Der

AU - Su, Hong Ren

AU - Cheng, Philip E.

AU - Liou, Michelle

AU - Aston, John A D

AU - Tsai, Arthur C.

AU - Chen, Cheng Yu

PY - 2009/6

Y1 - 2009/6

N2 - This study proposes a segmentation method for brain MR images using a distribution transformation approach. The method extends traditional Gaussian mixtures expectation-maximization segmentation to a power transformed version of mixed intensity distributions, which includes Gaussian mixtures as a special case. As MR intensities tend to exhibit non-Gaussianity due to partial volume effects, the proposed method is designed to fit non-Gaussian tissue intensity distributions. One advantage of the method is that it is intuitively appealing and computationally simple. To avoid performance degradation caused by intensity inhomogeneity, different methods for correcting bias fields were applied prior to image segmentation, and their correction effects on the segmentation results were examined in the empirical study. The partitions of brain tissues (i.e., gray and white matter) resulting from the method were validated and evaluated against manual segmentation results based on 38 real T1-weighted image volumes from the internet brain segmentation repository, and 18 simulated image volumes from Brain Web. The Jaccard and Dice similarity indexes were computed to evaluate the performance of the proposed approach relative to the expert segmentations. Empirical results suggested that the proposed segmentation method yielded higher similarity measures for both gray matter and white matter as compared with those based on the traditional segmentation using the Gaussian mixtures approach.

AB - This study proposes a segmentation method for brain MR images using a distribution transformation approach. The method extends traditional Gaussian mixtures expectation-maximization segmentation to a power transformed version of mixed intensity distributions, which includes Gaussian mixtures as a special case. As MR intensities tend to exhibit non-Gaussianity due to partial volume effects, the proposed method is designed to fit non-Gaussian tissue intensity distributions. One advantage of the method is that it is intuitively appealing and computationally simple. To avoid performance degradation caused by intensity inhomogeneity, different methods for correcting bias fields were applied prior to image segmentation, and their correction effects on the segmentation results were examined in the empirical study. The partitions of brain tissues (i.e., gray and white matter) resulting from the method were validated and evaluated against manual segmentation results based on 38 real T1-weighted image volumes from the internet brain segmentation repository, and 18 simulated image volumes from Brain Web. The Jaccard and Dice similarity indexes were computed to evaluate the performance of the proposed approach relative to the expert segmentations. Empirical results suggested that the proposed segmentation method yielded higher similarity measures for both gray matter and white matter as compared with those based on the traditional segmentation using the Gaussian mixtures approach.

KW - Box-cox transformation

KW - Expectation-maximization (EM) algorithm

KW - Gaussian mixtures

KW - Statistical segmentation

UR - http://www.scopus.com/inward/record.url?scp=66249108273&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=66249108273&partnerID=8YFLogxK

U2 - 10.1109/TMI.2009.2012896

DO - 10.1109/TMI.2009.2012896

M3 - Article

C2 - 19164075

AN - SCOPUS:66249108273

VL - 28

SP - 894

EP - 905

JO - IEEE Transactions on Medical Imaging

JF - IEEE Transactions on Medical Imaging

SN - 0278-0062

IS - 6

M1 - 4752750

ER -