Integration of sparse multi-modality representation and anatomical constraint for isointense infant brain MR image segmentation

Li Wang, Feng Shi, Yaozong Gao, Gang Li, John H. Gilmore, Weili Lin, Dinggang Shen

Research output: Contribution to journalArticle

70 Citations (Scopus)

Abstract

Segmentation of infant brain MR images is challenging due to poor spatial resolution, severe partial volume effect, and the ongoing maturation and myelination processes. During the first year of life, the brain image contrast between white and gray matters undergoes dramatic changes. In particular, the image contrast inverses around 6-8. months of age, where the white and gray matter tissues are isointense in T1 and T2 weighted images and hence exhibit the extremely low tissue contrast, posing significant challenges for automated segmentation. In this paper, we propose a general framework that adopts sparse representation to fuse the multi-modality image information and further incorporate the anatomical constraints for brain tissue segmentation. Specifically, we first derive an initial segmentation from a library of aligned images with ground-truth segmentations by using sparse representation in a patch-based fashion for the multi-modality T1, T2 and FA images. The segmentation result is further iteratively refined by integration of the anatomical constraint. The proposed method was evaluated on 22 infant brain MR images acquired at around 6. months of age by using a leave-one-out cross-validation, as well as other 10 unseen testing subjects. Our method achieved a high accuracy for the Dice ratios that measure the volume overlap between automated and manual segmentations, i.e., 0.889 ± 0.008 for white matter and 0.870 ± 0.006 for gray matter.

Original languageEnglish
Pages (from-to)152-164
Number of pages13
JournalNeuroImage
Volume89
DOIs
Publication statusPublished - 2014 Apr 1

Fingerprint

Brain
Libraries
White Matter
Gray Matter

Keywords

  • Anatomical constraint
  • Infant brain images
  • Isointense stage
  • Low contrast
  • Sparse representation
  • Tissue segmentation

ASJC Scopus subject areas

  • Cognitive Neuroscience
  • Neurology

Cite this

Integration of sparse multi-modality representation and anatomical constraint for isointense infant brain MR image segmentation. / Wang, Li; Shi, Feng; Gao, Yaozong; Li, Gang; Gilmore, John H.; Lin, Weili; Shen, Dinggang.

In: NeuroImage, Vol. 89, 01.04.2014, p. 152-164.

Research output: Contribution to journalArticle

Wang, Li ; Shi, Feng ; Gao, Yaozong ; Li, Gang ; Gilmore, John H. ; Lin, Weili ; Shen, Dinggang. / Integration of sparse multi-modality representation and anatomical constraint for isointense infant brain MR image segmentation. In: NeuroImage. 2014 ; Vol. 89. pp. 152-164.
@article{4da68434c12a48b0896773edd2a7c2ca,
title = "Integration of sparse multi-modality representation and anatomical constraint for isointense infant brain MR image segmentation",
abstract = "Segmentation of infant brain MR images is challenging due to poor spatial resolution, severe partial volume effect, and the ongoing maturation and myelination processes. During the first year of life, the brain image contrast between white and gray matters undergoes dramatic changes. In particular, the image contrast inverses around 6-8. months of age, where the white and gray matter tissues are isointense in T1 and T2 weighted images and hence exhibit the extremely low tissue contrast, posing significant challenges for automated segmentation. In this paper, we propose a general framework that adopts sparse representation to fuse the multi-modality image information and further incorporate the anatomical constraints for brain tissue segmentation. Specifically, we first derive an initial segmentation from a library of aligned images with ground-truth segmentations by using sparse representation in a patch-based fashion for the multi-modality T1, T2 and FA images. The segmentation result is further iteratively refined by integration of the anatomical constraint. The proposed method was evaluated on 22 infant brain MR images acquired at around 6. months of age by using a leave-one-out cross-validation, as well as other 10 unseen testing subjects. Our method achieved a high accuracy for the Dice ratios that measure the volume overlap between automated and manual segmentations, i.e., 0.889 ± 0.008 for white matter and 0.870 ± 0.006 for gray matter.",
keywords = "Anatomical constraint, Infant brain images, Isointense stage, Low contrast, Sparse representation, Tissue segmentation",
author = "Li Wang and Feng Shi and Yaozong Gao and Gang Li and Gilmore, {John H.} and Weili Lin and Dinggang Shen",
year = "2014",
month = "4",
day = "1",
doi = "10.1016/j.neuroimage.2013.11.040",
language = "English",
volume = "89",
pages = "152--164",
journal = "NeuroImage",
issn = "1053-8119",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Integration of sparse multi-modality representation and anatomical constraint for isointense infant brain MR image segmentation

AU - Wang, Li

AU - Shi, Feng

AU - Gao, Yaozong

AU - Li, Gang

AU - Gilmore, John H.

AU - Lin, Weili

AU - Shen, Dinggang

PY - 2014/4/1

Y1 - 2014/4/1

N2 - Segmentation of infant brain MR images is challenging due to poor spatial resolution, severe partial volume effect, and the ongoing maturation and myelination processes. During the first year of life, the brain image contrast between white and gray matters undergoes dramatic changes. In particular, the image contrast inverses around 6-8. months of age, where the white and gray matter tissues are isointense in T1 and T2 weighted images and hence exhibit the extremely low tissue contrast, posing significant challenges for automated segmentation. In this paper, we propose a general framework that adopts sparse representation to fuse the multi-modality image information and further incorporate the anatomical constraints for brain tissue segmentation. Specifically, we first derive an initial segmentation from a library of aligned images with ground-truth segmentations by using sparse representation in a patch-based fashion for the multi-modality T1, T2 and FA images. The segmentation result is further iteratively refined by integration of the anatomical constraint. The proposed method was evaluated on 22 infant brain MR images acquired at around 6. months of age by using a leave-one-out cross-validation, as well as other 10 unseen testing subjects. Our method achieved a high accuracy for the Dice ratios that measure the volume overlap between automated and manual segmentations, i.e., 0.889 ± 0.008 for white matter and 0.870 ± 0.006 for gray matter.

AB - Segmentation of infant brain MR images is challenging due to poor spatial resolution, severe partial volume effect, and the ongoing maturation and myelination processes. During the first year of life, the brain image contrast between white and gray matters undergoes dramatic changes. In particular, the image contrast inverses around 6-8. months of age, where the white and gray matter tissues are isointense in T1 and T2 weighted images and hence exhibit the extremely low tissue contrast, posing significant challenges for automated segmentation. In this paper, we propose a general framework that adopts sparse representation to fuse the multi-modality image information and further incorporate the anatomical constraints for brain tissue segmentation. Specifically, we first derive an initial segmentation from a library of aligned images with ground-truth segmentations by using sparse representation in a patch-based fashion for the multi-modality T1, T2 and FA images. The segmentation result is further iteratively refined by integration of the anatomical constraint. The proposed method was evaluated on 22 infant brain MR images acquired at around 6. months of age by using a leave-one-out cross-validation, as well as other 10 unseen testing subjects. Our method achieved a high accuracy for the Dice ratios that measure the volume overlap between automated and manual segmentations, i.e., 0.889 ± 0.008 for white matter and 0.870 ± 0.006 for gray matter.

KW - Anatomical constraint

KW - Infant brain images

KW - Isointense stage

KW - Low contrast

KW - Sparse representation

KW - Tissue segmentation

UR - http://www.scopus.com/inward/record.url?scp=84892862607&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84892862607&partnerID=8YFLogxK

U2 - 10.1016/j.neuroimage.2013.11.040

DO - 10.1016/j.neuroimage.2013.11.040

M3 - Article

VL - 89

SP - 152

EP - 164

JO - NeuroImage

JF - NeuroImage

SN - 1053-8119

ER -