Combining generative models for multifocal glioma segmentation and registration

Dongjin Kwon, Russell T. Shinohara, Hamed Akbari, Christos Davatzikos

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

Abstract

In this paper, we propose a new method for simultaneously segmenting brain scans of glioma patients and registering these scans to a normal atlas. Performing joint segmentation and registration for brain tumors is very challenging when tumors include multifocal masses and have complex shapes with heterogeneous textures. Our approach grows tumors for each mass from multiple seed points using a tumor growth model and modifies a normal atlas into one with tumors and edema using the combined results of grown tumors. We also generate a tumor shape prior via the random walk with restart, utilizing multiple tumor seeds as initial foreground information. We then incorporate this shape prior into an EM framework which estimates the mapping between the modified atlas and the scans, posteriors for each tissue labels, and the tumor growth model parameters. We apply our method to the BRATS 2013 leaderboard dataset to evaluate segmentation performance. Our method shows the best performance among all participants.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages763-770
Number of pages8
Volume8673 LNCS
EditionPART 1
ISBN (Print)9783319104034
DOIs
StatePublished - 2014
Externally publishedYes
Event17th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2014 - Boston, MA, United States
Duration: Sep 14 2014Sep 18 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume8673 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other17th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2014
CountryUnited States
CityBoston, MA
Period9/14/149/18/14

Fingerprint

Generative Models
Registration
Tumors
Tumor
Segmentation
Atlas
Tumor Growth
Growth Model
Brain Tumor
Seed
Restart
Brain
Texture
Random walk
Labels
Evaluate
Textures
Estimate
Tissue

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Kwon, D., Shinohara, R. T., Akbari, H., & Davatzikos, C. (2014). Combining generative models for multifocal glioma segmentation and registration. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 1 ed., Vol. 8673 LNCS, pp. 763-770). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8673 LNCS, No. PART 1). Springer Verlag. https://doi.org/10.1007/978-3-319-10404-1_95

Combining generative models for multifocal glioma segmentation and registration. / Kwon, Dongjin; Shinohara, Russell T.; Akbari, Hamed; Davatzikos, Christos.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8673 LNCS PART 1. ed. Springer Verlag, 2014. p. 763-770 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8673 LNCS, No. PART 1).

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

Kwon, D, Shinohara, RT, Akbari, H & Davatzikos, C 2014, Combining generative models for multifocal glioma segmentation and registration. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 edn, vol. 8673 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 1, vol. 8673 LNCS, Springer Verlag, pp. 763-770, 17th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2014, Boston, MA, United States, 9/14/14. https://doi.org/10.1007/978-3-319-10404-1_95
Kwon D, Shinohara RT, Akbari H, Davatzikos C. Combining generative models for multifocal glioma segmentation and registration. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 ed. Vol. 8673 LNCS. Springer Verlag. 2014. p. 763-770. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1). https://doi.org/10.1007/978-3-319-10404-1_95
Kwon, Dongjin ; Shinohara, Russell T. ; Akbari, Hamed ; Davatzikos, Christos. / Combining generative models for multifocal glioma segmentation and registration. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8673 LNCS PART 1. ed. Springer Verlag, 2014. pp. 763-770 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
@inproceedings{586fbf9b237b4ff2b85e7ddd50f4d5c5,
title = "Combining generative models for multifocal glioma segmentation and registration",
abstract = "In this paper, we propose a new method for simultaneously segmenting brain scans of glioma patients and registering these scans to a normal atlas. Performing joint segmentation and registration for brain tumors is very challenging when tumors include multifocal masses and have complex shapes with heterogeneous textures. Our approach grows tumors for each mass from multiple seed points using a tumor growth model and modifies a normal atlas into one with tumors and edema using the combined results of grown tumors. We also generate a tumor shape prior via the random walk with restart, utilizing multiple tumor seeds as initial foreground information. We then incorporate this shape prior into an EM framework which estimates the mapping between the modified atlas and the scans, posteriors for each tissue labels, and the tumor growth model parameters. We apply our method to the BRATS 2013 leaderboard dataset to evaluate segmentation performance. Our method shows the best performance among all participants.",
author = "Dongjin Kwon and Shinohara, {Russell T.} and Hamed Akbari and Christos Davatzikos",
year = "2014",
doi = "10.1007/978-3-319-10404-1_95",
language = "English (US)",
isbn = "9783319104034",
volume = "8673 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
number = "PART 1",
pages = "763--770",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 1",

}

TY - GEN

T1 - Combining generative models for multifocal glioma segmentation and registration

AU - Kwon, Dongjin

AU - Shinohara, Russell T.

AU - Akbari, Hamed

AU - Davatzikos, Christos

PY - 2014

Y1 - 2014

N2 - In this paper, we propose a new method for simultaneously segmenting brain scans of glioma patients and registering these scans to a normal atlas. Performing joint segmentation and registration for brain tumors is very challenging when tumors include multifocal masses and have complex shapes with heterogeneous textures. Our approach grows tumors for each mass from multiple seed points using a tumor growth model and modifies a normal atlas into one with tumors and edema using the combined results of grown tumors. We also generate a tumor shape prior via the random walk with restart, utilizing multiple tumor seeds as initial foreground information. We then incorporate this shape prior into an EM framework which estimates the mapping between the modified atlas and the scans, posteriors for each tissue labels, and the tumor growth model parameters. We apply our method to the BRATS 2013 leaderboard dataset to evaluate segmentation performance. Our method shows the best performance among all participants.

AB - In this paper, we propose a new method for simultaneously segmenting brain scans of glioma patients and registering these scans to a normal atlas. Performing joint segmentation and registration for brain tumors is very challenging when tumors include multifocal masses and have complex shapes with heterogeneous textures. Our approach grows tumors for each mass from multiple seed points using a tumor growth model and modifies a normal atlas into one with tumors and edema using the combined results of grown tumors. We also generate a tumor shape prior via the random walk with restart, utilizing multiple tumor seeds as initial foreground information. We then incorporate this shape prior into an EM framework which estimates the mapping between the modified atlas and the scans, posteriors for each tissue labels, and the tumor growth model parameters. We apply our method to the BRATS 2013 leaderboard dataset to evaluate segmentation performance. Our method shows the best performance among all participants.

UR - http://www.scopus.com/inward/record.url?scp=84906971511&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84906971511&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-10404-1_95

DO - 10.1007/978-3-319-10404-1_95

M3 - Conference contribution

SN - 9783319104034

VL - 8673 LNCS

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

SP - 763

EP - 770

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

PB - Springer Verlag

ER -