Multi-object geodesic active contours (MOGAC)

Blake C. Lucas; Michael Kazhdan; Russell H. Taylor

doi:10.1007/978-3-642-33418-4_50

Multi-object geodesic active contours (MOGAC)

Blake C. Lucas, Michael Kazhdan, Russell H. Taylor

Whiting School of Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

9 Scopus citations

Abstract

An emerging topic is to build image segmentation systems that can segment hundreds to thousands of objects (i.e. cell segmentation \ tracking, full brain parcellation, full body segmentation, etc.). Multi-object Level Set Methods (MLSM) perform this task with the benefit of sub-pixel precision. However, current implementations of MLSM are not as computationally or memory efficient as their region growing and graph cut counterparts which lack sub-pixel precision. To address this performance gap, we present a novel parallel implementation of MLSM that leverages the sparse properties of the algorithm to minimize its memory footprint for multiple objects. The new method, Multi-Object Geodesic Active Contours (MOGAC), can represent N objects with just two functions: a label mask image and unsigned distance field. The time complexity of the algorithm is shown to be O((M^d)/P) for M^d pixels and P processing units in dimension d={2,3}, independent of the number of objects. Results are presented for 2D and 3D image segmentation problems.

Original language	English (US)
Title of host publication	Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings
Editors	Albert Montillo, Bjoern H. Menze, Le Lu, Georg Langs, Antonio Criminisi, Bjoern H. Menze, Nicholas Ayache, Hervé Delingette, Zhuowen Tu, Georg Langs, Polina Golland, Kensaku Mori
Publisher	Springer Verlag
Pages	404-412
Number of pages	9
ISBN (Print)	9783642334177
DOIs	https://doi.org/10.1007/978-3-642-33418-4_50
State	Published - 2012
Event	15th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2012 - Nice, France Duration: Oct 5 2012 → Oct 5 2012

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	7511 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	15th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2012
Country/Territory	France
City	Nice
Period	10/5/12 → 10/5/12

Keywords

Active contours
Level set
Parallel
Segmentation

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-642-33418-4_50

Cite this

Lucas, B. C., Kazhdan, M., & Taylor, R. H. (2012). Multi-object geodesic active contours (MOGAC). In A. Montillo, B. H. Menze, L. Lu, G. Langs, A. Criminisi, B. H. Menze, N. Ayache, H. Delingette, Z. Tu, G. Langs, P. Golland, & K. Mori (Eds.), Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings (pp. 404-412). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7511 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-642-33418-4_50

Multi-object geodesic active contours (MOGAC). / Lucas, Blake C.; Kazhdan, Michael; Taylor, Russell H.
Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings. ed. / Albert Montillo; Bjoern H. Menze; Le Lu; Georg Langs; Antonio Criminisi; Bjoern H. Menze; Nicholas Ayache; Hervé Delingette; Zhuowen Tu; Georg Langs; Polina Golland; Kensaku Mori. Springer Verlag, 2012. p. 404-412 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7511 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Lucas, BC, Kazhdan, M & Taylor, RH 2012, Multi-object geodesic active contours (MOGAC). in A Montillo, BH Menze, L Lu, G Langs, A Criminisi, BH Menze, N Ayache, H Delingette, Z Tu, G Langs, P Golland & K Mori (eds), Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7511 LNCS, Springer Verlag, pp. 404-412, 15th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2012, Nice, France, 10/5/12. https://doi.org/10.1007/978-3-642-33418-4_50

Lucas BC, Kazhdan M, Taylor RH. Multi-object geodesic active contours (MOGAC). In Montillo A, Menze BH, Lu L, Langs G, Criminisi A, Menze BH, Ayache N, Delingette H, Tu Z, Langs G, Golland P, Mori K, editors, Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings. Springer Verlag. 2012. p. 404-412. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-33418-4_50

Lucas, Blake C. ; Kazhdan, Michael ; Taylor, Russell H. / Multi-object geodesic active contours (MOGAC). Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings. editor / Albert Montillo ; Bjoern H. Menze ; Le Lu ; Georg Langs ; Antonio Criminisi ; Bjoern H. Menze ; Nicholas Ayache ; Hervé Delingette ; Zhuowen Tu ; Georg Langs ; Polina Golland ; Kensaku Mori. Springer Verlag, 2012. pp. 404-412 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{d98b3d6a65ac425bb5bd9f087300ebbb,

title = "Multi-object geodesic active contours (MOGAC)",

abstract = "An emerging topic is to build image segmentation systems that can segment hundreds to thousands of objects (i.e. cell segmentation \ tracking, full brain parcellation, full body segmentation, etc.). Multi-object Level Set Methods (MLSM) perform this task with the benefit of sub-pixel precision. However, current implementations of MLSM are not as computationally or memory efficient as their region growing and graph cut counterparts which lack sub-pixel precision. To address this performance gap, we present a novel parallel implementation of MLSM that leverages the sparse properties of the algorithm to minimize its memory footprint for multiple objects. The new method, Multi-Object Geodesic Active Contours (MOGAC), can represent N objects with just two functions: a label mask image and unsigned distance field. The time complexity of the algorithm is shown to be O((M^d)/P) for M^d pixels and P processing units in dimension d={2,3}, independent of the number of objects. Results are presented for 2D and 3D image segmentation problems.",

keywords = "Active contours, Level set, Parallel, Segmentation",

author = "Lucas, {Blake C.} and Michael Kazhdan and Taylor, {Russell H.}",

note = "Publisher Copyright: {\textcopyright} Springer-Verlag Berlin Heidelberg 2012.; 15th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2012 ; Conference date: 05-10-2012 Through 05-10-2012",

year = "2012",

doi = "10.1007/978-3-642-33418-4_50",

language = "English (US)",

isbn = "9783642334177",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "404--412",

editor = "Albert Montillo and Menze, {Bjoern H.} and Le Lu and Georg Langs and Antonio Criminisi and Menze, {Bjoern H.} and Nicholas Ayache and Herv{\'e} Delingette and Zhuowen Tu and Georg Langs and Polina Golland and Kensaku Mori",

booktitle = "Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings",

}

TY - GEN

T1 - Multi-object geodesic active contours (MOGAC)

AU - Lucas, Blake C.

AU - Kazhdan, Michael

AU - Taylor, Russell H.

N1 - Publisher Copyright: © Springer-Verlag Berlin Heidelberg 2012.

PY - 2012

Y1 - 2012

N2 - An emerging topic is to build image segmentation systems that can segment hundreds to thousands of objects (i.e. cell segmentation \ tracking, full brain parcellation, full body segmentation, etc.). Multi-object Level Set Methods (MLSM) perform this task with the benefit of sub-pixel precision. However, current implementations of MLSM are not as computationally or memory efficient as their region growing and graph cut counterparts which lack sub-pixel precision. To address this performance gap, we present a novel parallel implementation of MLSM that leverages the sparse properties of the algorithm to minimize its memory footprint for multiple objects. The new method, Multi-Object Geodesic Active Contours (MOGAC), can represent N objects with just two functions: a label mask image and unsigned distance field. The time complexity of the algorithm is shown to be O((M^d)/P) for M^d pixels and P processing units in dimension d={2,3}, independent of the number of objects. Results are presented for 2D and 3D image segmentation problems.

AB - An emerging topic is to build image segmentation systems that can segment hundreds to thousands of objects (i.e. cell segmentation \ tracking, full brain parcellation, full body segmentation, etc.). Multi-object Level Set Methods (MLSM) perform this task with the benefit of sub-pixel precision. However, current implementations of MLSM are not as computationally or memory efficient as their region growing and graph cut counterparts which lack sub-pixel precision. To address this performance gap, we present a novel parallel implementation of MLSM that leverages the sparse properties of the algorithm to minimize its memory footprint for multiple objects. The new method, Multi-Object Geodesic Active Contours (MOGAC), can represent N objects with just two functions: a label mask image and unsigned distance field. The time complexity of the algorithm is shown to be O((M^d)/P) for M^d pixels and P processing units in dimension d={2,3}, independent of the number of objects. Results are presented for 2D and 3D image segmentation problems.

KW - Active contours

KW - Level set

KW - Parallel

KW - Segmentation

UR - http://www.scopus.com/inward/record.url?scp=84872900369&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84872900369&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-33418-4_50

DO - 10.1007/978-3-642-33418-4_50

M3 - Conference contribution

C2 - 23286074

AN - SCOPUS:84872900369

SN - 9783642334177

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

SP - 404

EP - 412

BT - Medical Image Computing and Computer-Assisted Intervention - MICCAI2012 - 15th International Conference, Proceedings

A2 - Montillo, Albert

A2 - Menze, Bjoern H.

A2 - Lu, Le

A2 - Langs, Georg

A2 - Criminisi, Antonio

A2 - Menze, Bjoern H.

A2 - Ayache, Nicholas

A2 - Delingette, Hervé

A2 - Tu, Zhuowen

A2 - Langs, Georg

A2 - Golland, Polina

A2 - Mori, Kensaku

PB - Springer Verlag

T2 - 15th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2012

Y2 - 5 October 2012 through 5 October 2012

ER -

Multi-object geodesic active contours (MOGAC)

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this