A cross-center smoothness prior for variational Bayesian brain tissue segmentation

Wouter M. Kouw, Silas N. Ørting, Jens Petersen, Kim S. Pedersen, Marleen de Bruijne

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

Suppose one is faced with the challenge of tissue segmentation in MR images, without annotators at their center to provide labeled training data. One option is to go to another medical center for a trained classifier. Sadly, tissue classifiers do not generalize well across centers due to voxel intensity shifts caused by center-specific acquisition protocols. However, certain aspects of segmentations, such as spatial smoothness, remain relatively consistent and can be learned separately. Here we present a smoothness prior that is fit to segmentations produced at another medical center. This informative prior is presented to an unsupervised Bayesian model. The model clusters the voxel intensities, such that it produces segmentations that are similarly smooth to those of the other medical center. In addition, the unsupervised Bayesian model is extended to a semi-supervised variant, which needs no visual interpretation of clusters into tissues.

LanguageEnglish
Title of host publicationInternational Conference on Information Processing in Medical Imaging
EditorsSiqi Bao, Albert C.S. Chung, James C. Gee, Paul A. Yushkevich
Place of PublicationCham
PublisherSpringer
Pages360-371
Number of pages12
ISBN (Electronic)978-3-030-20351-1
ISBN (Print)978-3-030-20350-4
DOIs
StatePublished - 1 Jan 2019
Externally publishedYes
Event26th International Conference on Information Processing in Medical Imaging, IPMI 2019 - Hong Kong, China
Duration: 2 Jun 20197 Jun 2019

Publication series

NameLecture Notes in Computer Science
Volume11492 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference26th International Conference on Information Processing in Medical Imaging, IPMI 2019
CountryChina
CityHong Kong
Period2/06/197/06/19

Fingerprint

Brain
Tissue
Classifiers

Keywords

  • Bayesian transfer learning
  • Image segmentation
  • Variational inference

Cite this

Kouw, W. M., Ørting, S. N., Petersen, J., Pedersen, K. S., & de Bruijne, M. (2019). A cross-center smoothness prior for variational Bayesian brain tissue segmentation. In S. Bao, A. C. S. Chung, J. C. Gee, & P. A. Yushkevich (Eds.), International Conference on Information Processing in Medical Imaging (pp. 360-371). (Lecture Notes in Computer Science; Vol. 11492 LNCS). Cham: Springer. DOI: 10.1007/978-3-030-20351-1_27
Kouw, Wouter M. ; Ørting, Silas N. ; Petersen, Jens ; Pedersen, Kim S. ; de Bruijne, Marleen. / A cross-center smoothness prior for variational Bayesian brain tissue segmentation. International Conference on Information Processing in Medical Imaging. editor / Siqi Bao ; Albert C.S. Chung ; James C. Gee ; Paul A. Yushkevich. Cham : Springer, 2019. pp. 360-371 (Lecture Notes in Computer Science).
@inproceedings{bddd9567e93e4c6d969cfae67e69355c,
title = "A cross-center smoothness prior for variational Bayesian brain tissue segmentation",
abstract = "Suppose one is faced with the challenge of tissue segmentation in MR images, without annotators at their center to provide labeled training data. One option is to go to another medical center for a trained classifier. Sadly, tissue classifiers do not generalize well across centers due to voxel intensity shifts caused by center-specific acquisition protocols. However, certain aspects of segmentations, such as spatial smoothness, remain relatively consistent and can be learned separately. Here we present a smoothness prior that is fit to segmentations produced at another medical center. This informative prior is presented to an unsupervised Bayesian model. The model clusters the voxel intensities, such that it produces segmentations that are similarly smooth to those of the other medical center. In addition, the unsupervised Bayesian model is extended to a semi-supervised variant, which needs no visual interpretation of clusters into tissues.",
keywords = "Bayesian transfer learning, Image segmentation, Variational inference",
author = "Kouw, {Wouter M.} and {\O}rting, {Silas N.} and Jens Petersen and Pedersen, {Kim S.} and {de Bruijne}, Marleen",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/978-3-030-20351-1_27",
language = "English",
isbn = "978-3-030-20350-4",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "360--371",
editor = "Siqi Bao and Chung, {Albert C.S.} and Gee, {James C.} and Yushkevich, {Paul A.}",
booktitle = "International Conference on Information Processing in Medical Imaging",
address = "Germany",

}

Kouw, WM, Ørting, SN, Petersen, J, Pedersen, KS & de Bruijne, M 2019, A cross-center smoothness prior for variational Bayesian brain tissue segmentation. in S Bao, ACS Chung, JC Gee & PA Yushkevich (eds), International Conference on Information Processing in Medical Imaging. Lecture Notes in Computer Science, vol. 11492 LNCS, Springer, Cham, pp. 360-371, 26th International Conference on Information Processing in Medical Imaging, IPMI 2019, Hong Kong, China, 2/06/19. DOI: 10.1007/978-3-030-20351-1_27

A cross-center smoothness prior for variational Bayesian brain tissue segmentation. / Kouw, Wouter M.; Ørting, Silas N.; Petersen, Jens; Pedersen, Kim S.; de Bruijne, Marleen.

International Conference on Information Processing in Medical Imaging. ed. / Siqi Bao; Albert C.S. Chung; James C. Gee; Paul A. Yushkevich. Cham : Springer, 2019. p. 360-371 (Lecture Notes in Computer Science; Vol. 11492 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - A cross-center smoothness prior for variational Bayesian brain tissue segmentation

AU - Kouw,Wouter M.

AU - Ørting,Silas N.

AU - Petersen,Jens

AU - Pedersen,Kim S.

AU - de Bruijne,Marleen

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Suppose one is faced with the challenge of tissue segmentation in MR images, without annotators at their center to provide labeled training data. One option is to go to another medical center for a trained classifier. Sadly, tissue classifiers do not generalize well across centers due to voxel intensity shifts caused by center-specific acquisition protocols. However, certain aspects of segmentations, such as spatial smoothness, remain relatively consistent and can be learned separately. Here we present a smoothness prior that is fit to segmentations produced at another medical center. This informative prior is presented to an unsupervised Bayesian model. The model clusters the voxel intensities, such that it produces segmentations that are similarly smooth to those of the other medical center. In addition, the unsupervised Bayesian model is extended to a semi-supervised variant, which needs no visual interpretation of clusters into tissues.

AB - Suppose one is faced with the challenge of tissue segmentation in MR images, without annotators at their center to provide labeled training data. One option is to go to another medical center for a trained classifier. Sadly, tissue classifiers do not generalize well across centers due to voxel intensity shifts caused by center-specific acquisition protocols. However, certain aspects of segmentations, such as spatial smoothness, remain relatively consistent and can be learned separately. Here we present a smoothness prior that is fit to segmentations produced at another medical center. This informative prior is presented to an unsupervised Bayesian model. The model clusters the voxel intensities, such that it produces segmentations that are similarly smooth to those of the other medical center. In addition, the unsupervised Bayesian model is extended to a semi-supervised variant, which needs no visual interpretation of clusters into tissues.

KW - Bayesian transfer learning

KW - Image segmentation

KW - Variational inference

UR - http://www.scopus.com/inward/record.url?scp=85066129813&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-20351-1_27

DO - 10.1007/978-3-030-20351-1_27

M3 - Conference contribution

SN - 978-3-030-20350-4

T3 - Lecture Notes in Computer Science

SP - 360

EP - 371

BT - International Conference on Information Processing in Medical Imaging

PB - Springer

CY - Cham

ER -

Kouw WM, Ørting SN, Petersen J, Pedersen KS, de Bruijne M. A cross-center smoothness prior for variational Bayesian brain tissue segmentation. In Bao S, Chung ACS, Gee JC, Yushkevich PA, editors, International Conference on Information Processing in Medical Imaging. Cham: Springer. 2019. p. 360-371. (Lecture Notes in Computer Science). Available from, DOI: 10.1007/978-3-030-20351-1_27