Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images

Yuanjie Zheng, Sajjad Baloch, Sarah Englander, Mitchell D. Schnall, Dinggang Shen

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

Abstract

Accuracy of automatic cancer diagnosis is largely determined by two factors, namely, the precision of tumor segmentation, and the suitability of extracted features for discrimination between malignancy and benignancy. In this paper, we propose a new framework for accurate characterization of tumors in contrast enhanced MR images. First, a new graph cut based segmentation algorithm is developed for refining coarse manual segmentation, which allows precise identification of tumor regions. Second, by considering serial contrast-enhanced images as a single spatio-temporal image, a spatio-temporal model of segmented tumor is constructed to extract Spatio-Temporal Enhancement Patterns (STEPs). STEPs are designed to capture not only dynamic enhancement and architectural features, but also spatial variations of pixel-wise temporal enhancement of the tumor. While temporal enhancement features are extracted through Fourier transform, the resulting STEP framework captures spatial patterns of temporal enhancement features via moment invariants and rotation invariant Gabor textures. High accuracy of the proposed framework is a direct consequence of this two pronged approach, which is validated through experiments yielding, for instance, an area of 0.97 under the ROC curve.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages393-401
Number of pages9
Volume4792 LNCS
EditionPART 2
Publication statusPublished - 2007 Dec 1
Externally publishedYes
Event10th International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2007 - Brisbane, Australia
Duration: 2007 Oct 292007 Nov 2

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 2
Volume4792 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other10th International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2007
CountryAustralia
CityBrisbane
Period07/10/2907/11/2

Fingerprint

Tumors
Tumor
Segmentation
Enhancement
Breast Neoplasms
Neoplasms
Moment Invariants
Spatio-temporal Model
Refining
Rotation Invariant
Graph Cuts
Fourier transforms
Receiver Operating Characteristic Curve
Spatial Pattern
Textures
Pixels
Fourier Analysis
Discrimination
Texture
Fourier transform

ASJC Scopus subject areas

  • Computer Science(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Theoretical Computer Science

Cite this

Zheng, Y., Baloch, S., Englander, S., Schnall, M. D., & Shen, D. (2007). Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 2 ed., Vol. 4792 LNCS, pp. 393-401). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4792 LNCS, No. PART 2).

Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images. / Zheng, Yuanjie; Baloch, Sajjad; Englander, Sarah; Schnall, Mitchell D.; Shen, Dinggang.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4792 LNCS PART 2. ed. 2007. p. 393-401 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4792 LNCS, No. PART 2).

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

Zheng, Y, Baloch, S, Englander, S, Schnall, MD & Shen, D 2007, Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 2 edn, vol. 4792 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 4792 LNCS, pp. 393-401, 10th International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2007, Brisbane, Australia, 07/10/29.
Zheng Y, Baloch S, Englander S, Schnall MD, Shen D. Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 2 ed. Vol. 4792 LNCS. 2007. p. 393-401. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
Zheng, Yuanjie ; Baloch, Sajjad ; Englander, Sarah ; Schnall, Mitchell D. ; Shen, Dinggang. / Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4792 LNCS PART 2. ed. 2007. pp. 393-401 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
@inproceedings{6fc3e18430f74e1f92ee0c52444a8a8a,
title = "Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images",
abstract = "Accuracy of automatic cancer diagnosis is largely determined by two factors, namely, the precision of tumor segmentation, and the suitability of extracted features for discrimination between malignancy and benignancy. In this paper, we propose a new framework for accurate characterization of tumors in contrast enhanced MR images. First, a new graph cut based segmentation algorithm is developed for refining coarse manual segmentation, which allows precise identification of tumor regions. Second, by considering serial contrast-enhanced images as a single spatio-temporal image, a spatio-temporal model of segmented tumor is constructed to extract Spatio-Temporal Enhancement Patterns (STEPs). STEPs are designed to capture not only dynamic enhancement and architectural features, but also spatial variations of pixel-wise temporal enhancement of the tumor. While temporal enhancement features are extracted through Fourier transform, the resulting STEP framework captures spatial patterns of temporal enhancement features via moment invariants and rotation invariant Gabor textures. High accuracy of the proposed framework is a direct consequence of this two pronged approach, which is validated through experiments yielding, for instance, an area of 0.97 under the ROC curve.",
author = "Yuanjie Zheng and Sajjad Baloch and Sarah Englander and Schnall, {Mitchell D.} and Dinggang Shen",
year = "2007",
month = "12",
day = "1",
language = "English",
isbn = "9783540757580",
volume = "4792 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 2",
pages = "393--401",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 2",

}

TY - GEN

T1 - Segmentation and classification of breast tumor using dynamic contrast-enhanced MR images

AU - Zheng, Yuanjie

AU - Baloch, Sajjad

AU - Englander, Sarah

AU - Schnall, Mitchell D.

AU - Shen, Dinggang

PY - 2007/12/1

Y1 - 2007/12/1

N2 - Accuracy of automatic cancer diagnosis is largely determined by two factors, namely, the precision of tumor segmentation, and the suitability of extracted features for discrimination between malignancy and benignancy. In this paper, we propose a new framework for accurate characterization of tumors in contrast enhanced MR images. First, a new graph cut based segmentation algorithm is developed for refining coarse manual segmentation, which allows precise identification of tumor regions. Second, by considering serial contrast-enhanced images as a single spatio-temporal image, a spatio-temporal model of segmented tumor is constructed to extract Spatio-Temporal Enhancement Patterns (STEPs). STEPs are designed to capture not only dynamic enhancement and architectural features, but also spatial variations of pixel-wise temporal enhancement of the tumor. While temporal enhancement features are extracted through Fourier transform, the resulting STEP framework captures spatial patterns of temporal enhancement features via moment invariants and rotation invariant Gabor textures. High accuracy of the proposed framework is a direct consequence of this two pronged approach, which is validated through experiments yielding, for instance, an area of 0.97 under the ROC curve.

AB - Accuracy of automatic cancer diagnosis is largely determined by two factors, namely, the precision of tumor segmentation, and the suitability of extracted features for discrimination between malignancy and benignancy. In this paper, we propose a new framework for accurate characterization of tumors in contrast enhanced MR images. First, a new graph cut based segmentation algorithm is developed for refining coarse manual segmentation, which allows precise identification of tumor regions. Second, by considering serial contrast-enhanced images as a single spatio-temporal image, a spatio-temporal model of segmented tumor is constructed to extract Spatio-Temporal Enhancement Patterns (STEPs). STEPs are designed to capture not only dynamic enhancement and architectural features, but also spatial variations of pixel-wise temporal enhancement of the tumor. While temporal enhancement features are extracted through Fourier transform, the resulting STEP framework captures spatial patterns of temporal enhancement features via moment invariants and rotation invariant Gabor textures. High accuracy of the proposed framework is a direct consequence of this two pronged approach, which is validated through experiments yielding, for instance, an area of 0.97 under the ROC curve.

UR - http://www.scopus.com/inward/record.url?scp=38349137776&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=38349137776&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9783540757580

VL - 4792 LNCS

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

SP - 393

EP - 401

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

ER -