Brain-cloud: A generalized and flexible registration framework for brain MR images

Minjeong Kim, Guorong Wu, Qian Wang, Dinggang Shen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Image registration, which aligns a pair of fixed and moving images, is often tackled by the large shape and intensity variation between the images. As a remedy, we present a generalized registration framework that is capable to predict the initial deformation field between the fixed and moving images, even though their appearances are very different. For the prediction, we learn the prior knowledge on deformation from pre-observed images. Especially, our method is significantly differentiated from previous methods that are usually confined to a specific fixed image, to be flexible for handling arbitrary fixed and moving images. Specifically, our idea is to encapsulate many pre-observed images into a hierarchical infrastructure, termed as cloud, which is able to efficiently compute the deformation pathways between the pre-observed images. After anchoring the fixed and moving images to their respective port images (similar images in terms of intensity appearance) in the cloud, we predict the initial deformation between the fixed and moving images by the deformation pathway between the two port images. Thus, the remaining small deformation can be efficiently refined via most existing deformable registration methods. With the cloud, we have obtained promising registration results on both adult and infant brain images, demonstrating the advantage of the proposed registration framework in improving the registration performance.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages153-161
Number of pages9
Volume8090 LNCS
DOIs
Publication statusPublished - 2013 Dec 30
Externally publishedYes
Event6th International Workshop on Augmented Reality Environments for Medical Imaging and Computer-Assisted Interventions, MIAR 2013 and 8th International Workshop, AE-CAI 2013, Held in Conjunction with MICCAI 2013 - Nagoya, Japan
Duration: 2013 Sep 222013 Sep 22

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8090 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other6th International Workshop on Augmented Reality Environments for Medical Imaging and Computer-Assisted Interventions, MIAR 2013 and 8th International Workshop, AE-CAI 2013, Held in Conjunction with MICCAI 2013
CountryJapan
CityNagoya
Period13/9/2213/9/22

Fingerprint

Registration
Brain
Image registration
Framework
Pathway
Predict
Image Registration
Prior Knowledge
Infrastructure

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Kim, M., Wu, G., Wang, Q., & Shen, D. (2013). Brain-cloud: A generalized and flexible registration framework for brain MR images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8090 LNCS, pp. 153-161). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8090 LNCS). https://doi.org/10.1007/978-3-642-40843-4_17

Brain-cloud : A generalized and flexible registration framework for brain MR images. / Kim, Minjeong; Wu, Guorong; Wang, Qian; Shen, Dinggang.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8090 LNCS 2013. p. 153-161 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8090 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kim, M, Wu, G, Wang, Q & Shen, D 2013, Brain-cloud: A generalized and flexible registration framework for brain MR images. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 8090 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8090 LNCS, pp. 153-161, 6th International Workshop on Augmented Reality Environments for Medical Imaging and Computer-Assisted Interventions, MIAR 2013 and 8th International Workshop, AE-CAI 2013, Held in Conjunction with MICCAI 2013, Nagoya, Japan, 13/9/22. https://doi.org/10.1007/978-3-642-40843-4_17
Kim M, Wu G, Wang Q, Shen D. Brain-cloud: A generalized and flexible registration framework for brain MR images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8090 LNCS. 2013. p. 153-161. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-40843-4_17
Kim, Minjeong ; Wu, Guorong ; Wang, Qian ; Shen, Dinggang. / Brain-cloud : A generalized and flexible registration framework for brain MR images. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8090 LNCS 2013. pp. 153-161 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{2e1facda42b94fb3a3f9163c331c0b07,
title = "Brain-cloud: A generalized and flexible registration framework for brain MR images",
abstract = "Image registration, which aligns a pair of fixed and moving images, is often tackled by the large shape and intensity variation between the images. As a remedy, we present a generalized registration framework that is capable to predict the initial deformation field between the fixed and moving images, even though their appearances are very different. For the prediction, we learn the prior knowledge on deformation from pre-observed images. Especially, our method is significantly differentiated from previous methods that are usually confined to a specific fixed image, to be flexible for handling arbitrary fixed and moving images. Specifically, our idea is to encapsulate many pre-observed images into a hierarchical infrastructure, termed as cloud, which is able to efficiently compute the deformation pathways between the pre-observed images. After anchoring the fixed and moving images to their respective port images (similar images in terms of intensity appearance) in the cloud, we predict the initial deformation between the fixed and moving images by the deformation pathway between the two port images. Thus, the remaining small deformation can be efficiently refined via most existing deformable registration methods. With the cloud, we have obtained promising registration results on both adult and infant brain images, demonstrating the advantage of the proposed registration framework in improving the registration performance.",
author = "Minjeong Kim and Guorong Wu and Qian Wang and Dinggang Shen",
year = "2013",
month = "12",
day = "30",
doi = "10.1007/978-3-642-40843-4_17",
language = "English",
isbn = "9783642408427",
volume = "8090 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "153--161",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Brain-cloud

T2 - A generalized and flexible registration framework for brain MR images

AU - Kim, Minjeong

AU - Wu, Guorong

AU - Wang, Qian

AU - Shen, Dinggang

PY - 2013/12/30

Y1 - 2013/12/30

N2 - Image registration, which aligns a pair of fixed and moving images, is often tackled by the large shape and intensity variation between the images. As a remedy, we present a generalized registration framework that is capable to predict the initial deformation field between the fixed and moving images, even though their appearances are very different. For the prediction, we learn the prior knowledge on deformation from pre-observed images. Especially, our method is significantly differentiated from previous methods that are usually confined to a specific fixed image, to be flexible for handling arbitrary fixed and moving images. Specifically, our idea is to encapsulate many pre-observed images into a hierarchical infrastructure, termed as cloud, which is able to efficiently compute the deformation pathways between the pre-observed images. After anchoring the fixed and moving images to their respective port images (similar images in terms of intensity appearance) in the cloud, we predict the initial deformation between the fixed and moving images by the deformation pathway between the two port images. Thus, the remaining small deformation can be efficiently refined via most existing deformable registration methods. With the cloud, we have obtained promising registration results on both adult and infant brain images, demonstrating the advantage of the proposed registration framework in improving the registration performance.

AB - Image registration, which aligns a pair of fixed and moving images, is often tackled by the large shape and intensity variation between the images. As a remedy, we present a generalized registration framework that is capable to predict the initial deformation field between the fixed and moving images, even though their appearances are very different. For the prediction, we learn the prior knowledge on deformation from pre-observed images. Especially, our method is significantly differentiated from previous methods that are usually confined to a specific fixed image, to be flexible for handling arbitrary fixed and moving images. Specifically, our idea is to encapsulate many pre-observed images into a hierarchical infrastructure, termed as cloud, which is able to efficiently compute the deformation pathways between the pre-observed images. After anchoring the fixed and moving images to their respective port images (similar images in terms of intensity appearance) in the cloud, we predict the initial deformation between the fixed and moving images by the deformation pathway between the two port images. Thus, the remaining small deformation can be efficiently refined via most existing deformable registration methods. With the cloud, we have obtained promising registration results on both adult and infant brain images, demonstrating the advantage of the proposed registration framework in improving the registration performance.

UR - http://www.scopus.com/inward/record.url?scp=84890897353&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84890897353&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-40843-4_17

DO - 10.1007/978-3-642-40843-4_17

M3 - Conference contribution

AN - SCOPUS:84890897353

SN - 9783642408427

VL - 8090 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 153

EP - 161

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -