LINKS: Learning-based multi-source IntegratioN frameworK for Segmentation of infant brain images

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

Research output: Contribution to journalArticle

129 Citations (Scopus)

Abstract

Segmentation of infant brain MR images is challenging due to insufficient image quality, severe partial volume effect, and ongoing maturation and myelination processes. In the first year of life, the image contrast between white and gray matters of the infant brain undergoes dramatic changes. In particular, the image contrast is inverted around 6-8. months of age, and the white and gray matter tissues are isointense in both T1- and T2-weighted MR images and thus exhibit the extremely low tissue contrast, which poses significant challenges for automated segmentation. Most previous studies used multi-atlas label fusion strategy, which has the limitation of equally treating the different available image modalities and is often computationally expensive. To cope with these limitations, in this paper, we propose a novel learning-based multi-source integration framework for segmentation of infant brain images. Specifically, we employ the random forest technique to effectively integrate features from multi-source images together for tissue segmentation. Here, the multi-source images include initially only the multi-modality (T1, T2 and FA) images and later also the iteratively estimated and refined tissue probability maps of gray matter, white matter, and cerebrospinal fluid. Experimental results on 119 infants show that the proposed method achieves better performance than other state-of-the-art automated segmentation methods. Further validation was performed on the MICCAI grand challenge and the proposed method was ranked top among all competing methods. Moreover, to alleviate the possible anatomical errors, our method can also be combined with an anatomically-constrained multi-atlas labeling approach for further improving the segmentation accuracy.

Original languageEnglish
Pages (from-to)160-172
Number of pages13
JournalNeuroImage
Volume108
DOIs
Publication statusPublished - 2015 Mar 1

Fingerprint

Learning
Brain
Atlases
Cerebrospinal Fluid
White Matter
Gray Matter

Keywords

  • Context feature
  • Infant brain images
  • Isointense stage
  • Multi-modality
  • Random forest
  • Tissue segmentation

ASJC Scopus subject areas

  • Cognitive Neuroscience
  • Neurology

Cite this

LINKS : Learning-based multi-source IntegratioN frameworK for Segmentation of infant brain images. / Wang, Li; Gao, Yaozong; Shi, Feng; Li, Gang; Gilmore, John H.; Lin, Weili; Shen, Dinggang.

In: NeuroImage, Vol. 108, 01.03.2015, p. 160-172.

Research output: Contribution to journalArticle

Wang, Li ; Gao, Yaozong ; Shi, Feng ; Li, Gang ; Gilmore, John H. ; Lin, Weili ; Shen, Dinggang. / LINKS : Learning-based multi-source IntegratioN frameworK for Segmentation of infant brain images. In: NeuroImage. 2015 ; Vol. 108. pp. 160-172.
@article{f3d12fbc1e7f4ee7828f9ba85fd910c6,
title = "LINKS: Learning-based multi-source IntegratioN frameworK for Segmentation of infant brain images",
abstract = "Segmentation of infant brain MR images is challenging due to insufficient image quality, severe partial volume effect, and ongoing maturation and myelination processes. In the first year of life, the image contrast between white and gray matters of the infant brain undergoes dramatic changes. In particular, the image contrast is inverted around 6-8. months of age, and the white and gray matter tissues are isointense in both T1- and T2-weighted MR images and thus exhibit the extremely low tissue contrast, which poses significant challenges for automated segmentation. Most previous studies used multi-atlas label fusion strategy, which has the limitation of equally treating the different available image modalities and is often computationally expensive. To cope with these limitations, in this paper, we propose a novel learning-based multi-source integration framework for segmentation of infant brain images. Specifically, we employ the random forest technique to effectively integrate features from multi-source images together for tissue segmentation. Here, the multi-source images include initially only the multi-modality (T1, T2 and FA) images and later also the iteratively estimated and refined tissue probability maps of gray matter, white matter, and cerebrospinal fluid. Experimental results on 119 infants show that the proposed method achieves better performance than other state-of-the-art automated segmentation methods. Further validation was performed on the MICCAI grand challenge and the proposed method was ranked top among all competing methods. Moreover, to alleviate the possible anatomical errors, our method can also be combined with an anatomically-constrained multi-atlas labeling approach for further improving the segmentation accuracy.",
keywords = "Context feature, Infant brain images, Isointense stage, Multi-modality, Random forest, Tissue segmentation",
author = "Li Wang and Yaozong Gao and Feng Shi and Gang Li and Gilmore, {John H.} and Weili Lin and Dinggang Shen",
year = "2015",
month = "3",
day = "1",
doi = "10.1016/j.neuroimage.2014.12.042",
language = "English",
volume = "108",
pages = "160--172",
journal = "NeuroImage",
issn = "1053-8119",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - LINKS

T2 - Learning-based multi-source IntegratioN frameworK for Segmentation of infant brain images

AU - Wang, Li

AU - Gao, Yaozong

AU - Shi, Feng

AU - Li, Gang

AU - Gilmore, John H.

AU - Lin, Weili

AU - Shen, Dinggang

PY - 2015/3/1

Y1 - 2015/3/1

N2 - Segmentation of infant brain MR images is challenging due to insufficient image quality, severe partial volume effect, and ongoing maturation and myelination processes. In the first year of life, the image contrast between white and gray matters of the infant brain undergoes dramatic changes. In particular, the image contrast is inverted around 6-8. months of age, and the white and gray matter tissues are isointense in both T1- and T2-weighted MR images and thus exhibit the extremely low tissue contrast, which poses significant challenges for automated segmentation. Most previous studies used multi-atlas label fusion strategy, which has the limitation of equally treating the different available image modalities and is often computationally expensive. To cope with these limitations, in this paper, we propose a novel learning-based multi-source integration framework for segmentation of infant brain images. Specifically, we employ the random forest technique to effectively integrate features from multi-source images together for tissue segmentation. Here, the multi-source images include initially only the multi-modality (T1, T2 and FA) images and later also the iteratively estimated and refined tissue probability maps of gray matter, white matter, and cerebrospinal fluid. Experimental results on 119 infants show that the proposed method achieves better performance than other state-of-the-art automated segmentation methods. Further validation was performed on the MICCAI grand challenge and the proposed method was ranked top among all competing methods. Moreover, to alleviate the possible anatomical errors, our method can also be combined with an anatomically-constrained multi-atlas labeling approach for further improving the segmentation accuracy.

AB - Segmentation of infant brain MR images is challenging due to insufficient image quality, severe partial volume effect, and ongoing maturation and myelination processes. In the first year of life, the image contrast between white and gray matters of the infant brain undergoes dramatic changes. In particular, the image contrast is inverted around 6-8. months of age, and the white and gray matter tissues are isointense in both T1- and T2-weighted MR images and thus exhibit the extremely low tissue contrast, which poses significant challenges for automated segmentation. Most previous studies used multi-atlas label fusion strategy, which has the limitation of equally treating the different available image modalities and is often computationally expensive. To cope with these limitations, in this paper, we propose a novel learning-based multi-source integration framework for segmentation of infant brain images. Specifically, we employ the random forest technique to effectively integrate features from multi-source images together for tissue segmentation. Here, the multi-source images include initially only the multi-modality (T1, T2 and FA) images and later also the iteratively estimated and refined tissue probability maps of gray matter, white matter, and cerebrospinal fluid. Experimental results on 119 infants show that the proposed method achieves better performance than other state-of-the-art automated segmentation methods. Further validation was performed on the MICCAI grand challenge and the proposed method was ranked top among all competing methods. Moreover, to alleviate the possible anatomical errors, our method can also be combined with an anatomically-constrained multi-atlas labeling approach for further improving the segmentation accuracy.

KW - Context feature

KW - Infant brain images

KW - Isointense stage

KW - Multi-modality

KW - Random forest

KW - Tissue segmentation

UR - http://www.scopus.com/inward/record.url?scp=84920882082&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84920882082&partnerID=8YFLogxK

U2 - 10.1016/j.neuroimage.2014.12.042

DO - 10.1016/j.neuroimage.2014.12.042

M3 - Article

C2 - 25541188

AN - SCOPUS:84920882082

VL - 108

SP - 160

EP - 172

JO - NeuroImage

JF - NeuroImage

SN - 1053-8119

ER -