Image segmentation by shape particle filtering

Marleen de Bruijne; Mads Nielsen

doi:10.1109/ICPR.2004.1334630

Image segmentation by shape particle filtering

47 Citations (Scopus)

Abstract

Statistical appearance models are valuable tools in medical image segmentation. Current methods elegantly incorporate global shape and appearance, but cannot cope with local appearance variations and rely on an assumption of Gaussian gray value distribution. Furthermore, initialization near the optimal solution is required. We propose a shape inference method that is based on pixel classification, so that local and non-linear intensity variations are dealt with naturally, while a global shape model ensures a consistent segmentation. Optimization by stochastic sampling removes the need for accurate initialization. The method is demonstrated on vertebra segmentation in spine radiographs. Segmentation errors are below 2 mm in 88 out of 91 cases, with an average error of 1.4 mm.

Original language	English
Title of host publication	Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on
Volume	3
Publisher	IEEE Signal Processing Society
Publication date	2004
Pages	722- 725
ISBN (Print)	0-7695-2128-2
ISBN (Electronic)	1051-4651
DOIs	https://doi.org/10.1109/ICPR.2004.1334630
Publication status	Published - 2004
Externally published	Yes
Event	International Conference on Pattern Recognition (ICPR) - Cambridge, United Kingdom Duration: 29 Nov 2010 → … Conference number: 17

Conference

Conference	International Conference on Pattern Recognition (ICPR)
Number	17
Country/Territory	United Kingdom
City	Cambridge
Period	29/11/2010 → …

Access to Document

10.1109/ICPR.2004.1334630

http://ieeexplore.ieee.org/iel5/9258/29387/01334630.pdf?tp=&isnumber=&arnumber=1334630

Cite this

@inproceedings{f02ed060532811dd8d9f000ea68e967b,

title = "Image segmentation by shape particle filtering",

abstract = "Statistical appearance models are valuable tools in medical image segmentation. Current methods elegantly incorporate global shape and appearance, but cannot cope with local appearance variations and rely on an assumption of Gaussian gray value distribution. Furthermore, initialization near the optimal solution is required. We propose a shape inference method that is based on pixel classification, so that local and non-linear intensity variations are dealt with naturally, while a global shape model ensures a consistent segmentation. Optimization by stochastic sampling removes the need for accurate initialization. The method is demonstrated on vertebra segmentation in spine radiographs. Segmentation errors are below 2 mm in 88 out of 91 cases, with an average error of 1.4 mm.",

author = "{de Bruijne}, Marleen and Mads Nielsen",

year = "2004",

doi = "10.1109/ICPR.2004.1334630",

language = "English",

isbn = "0-7695-2128-2",

volume = "3",

pages = "722-- 725",

booktitle = "Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on",

publisher = "IEEE Signal Processing Society",

note = "International Conference on Pattern Recognition (ICPR) ; Conference date: 29-11-2010",

}

TY - GEN

T1 - Image segmentation by shape particle filtering

AU - de Bruijne, Marleen

AU - Nielsen, Mads

N1 - Conference code: 17

PY - 2004

Y1 - 2004

N2 - Statistical appearance models are valuable tools in medical image segmentation. Current methods elegantly incorporate global shape and appearance, but cannot cope with local appearance variations and rely on an assumption of Gaussian gray value distribution. Furthermore, initialization near the optimal solution is required. We propose a shape inference method that is based on pixel classification, so that local and non-linear intensity variations are dealt with naturally, while a global shape model ensures a consistent segmentation. Optimization by stochastic sampling removes the need for accurate initialization. The method is demonstrated on vertebra segmentation in spine radiographs. Segmentation errors are below 2 mm in 88 out of 91 cases, with an average error of 1.4 mm.

AB - Statistical appearance models are valuable tools in medical image segmentation. Current methods elegantly incorporate global shape and appearance, but cannot cope with local appearance variations and rely on an assumption of Gaussian gray value distribution. Furthermore, initialization near the optimal solution is required. We propose a shape inference method that is based on pixel classification, so that local and non-linear intensity variations are dealt with naturally, while a global shape model ensures a consistent segmentation. Optimization by stochastic sampling removes the need for accurate initialization. The method is demonstrated on vertebra segmentation in spine radiographs. Segmentation errors are below 2 mm in 88 out of 91 cases, with an average error of 1.4 mm.

U2 - 10.1109/ICPR.2004.1334630

DO - 10.1109/ICPR.2004.1334630

M3 - Article in proceedings

SN - 0-7695-2128-2

VL - 3

SP - 722

EP - 725

BT - Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on

PB - IEEE Signal Processing Society

T2 - International Conference on Pattern Recognition (ICPR)

Y2 - 29 November 2010

ER -

Image segmentation by shape particle filtering

Abstract

Conference

Access to Document

Fingerprint

Cite this