Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition

Adrian Galdran; Artzai Picón; Estibaliz Garrote; David Pardo

doi:10.1007/978-3-319-19390-8_66

Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition

Adrian Galdran^*
, Artzai Picón
, Estibaliz Garrote
, David Pardo

^*Corresponding author for this work

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

6 Citations (Scopus)

7 Downloads (Pure)

Abstract

Pectoral muscle segmentation on medio-lateral oblique views of mammograms represents an important preprocessing step in many mammographic image analysis tasks. Although its location can be perceptually obvious for a human observer, the variability in shape, size, and intensities of the pectoral muscle boundary turns its automatic segmentation into a challenging problem. In this work we propose to decompose the input mammogram into its textural and structural components at different scales prior to dynamically thresholding it into several levels. The resulting segmentations are refined with an active contour model and merged together by means of a simple voting scheme to remove possible outliers. Our method performs well compared to several other state-ofthe- art techniques. An average DICE similarity coefficient of 0.91 and mean Hausdorff distance of 3.66 ± 3.23 mm. validate our approach.

Original language	English
Title of host publication	Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings
Editors	Jaime S. Cardoso, Roberto Paredes, Xosé M. Pardo
Publisher	Springer Verlag
Pages	587-594
Number of pages	8
ISBN (Electronic)	9783319193892
ISBN (Print)	978-3-319-19390-8
DOIs	https://doi.org/10.1007/978-3-319-19390-8_66
Publication status	Published - 2015
Event	7th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2015 - Santiago de Compostela, Spain Duration: 17 Jun 2015 → 19 Jun 2015

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9117
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	7th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2015
Country/Territory	Spain
City	Santiago de Compostela
Period	17/06/15 → 19/06/15

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Breast cancer detection
Computer-aided diagnosis
Mammographic image analysis
Pectoral muscle segmentation

Access to Document

10.1007/978-3-319-19390-8_66

IbPRIA2015 Galdran camera readyFinal published version, 1.64 MB

Cite this

Galdran, A., Picón, A., Garrote, E., & Pardo, D. (2015). Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition. In J. S. Cardoso, R. Paredes, & X. M. Pardo (Eds.), Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings (pp. 587-594). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9117). Springer Verlag. https://doi.org/10.1007/978-3-319-19390-8_66

Galdran, Adrian ; Picón, Artzai ; Garrote, Estibaliz et al. / Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition. Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings. editor / Jaime S. Cardoso ; Roberto Paredes ; Xosé M. Pardo. Springer Verlag, 2015. pp. 587-594 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{1677e1c16ac84906b5e13101d4c9dc72,

title = "Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition",

abstract = "Pectoral muscle segmentation on medio-lateral oblique views of mammograms represents an important preprocessing step in many mammographic image analysis tasks. Although its location can be perceptually obvious for a human observer, the variability in shape, size, and intensities of the pectoral muscle boundary turns its automatic segmentation into a challenging problem. In this work we propose to decompose the input mammogram into its textural and structural components at different scales prior to dynamically thresholding it into several levels. The resulting segmentations are refined with an active contour model and merged together by means of a simple voting scheme to remove possible outliers. Our method performs well compared to several other state-ofthe- art techniques. An average DICE similarity coefficient of 0.91 and mean Hausdorff distance of 3.66 ± 3.23 mm. validate our approach.",

keywords = "Breast cancer detection, Computer-aided diagnosis, Mammographic image analysis, Pectoral muscle segmentation",

author = "Adrian Galdran and Artzai Pic{\'o}n and Estibaliz Garrote and David Pardo",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2015.; 7th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2015 ; Conference date: 17-06-2015 Through 19-06-2015",

year = "2015",

doi = "10.1007/978-3-319-19390-8\_66",

language = "English",

isbn = "978-3-319-19390-8",

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

publisher = "Springer Verlag",

pages = "587--594",

editor = "Cardoso, \{Jaime S.\} and Roberto Paredes and Pardo, \{Xos{\'e} M.\}",

booktitle = "Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings",

address = "Germany",

}

Galdran, A , Picón, A , Garrote, E & Pardo, D 2015, Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition. in JS Cardoso, R Paredes & XM Pardo (eds), Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9117, Springer Verlag, pp. 587-594, 7th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2015, Santiago de Compostela, Spain, 17/06/15. https://doi.org/10.1007/978-3-319-19390-8_66

Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition. / Galdran, Adrian ; Picón, Artzai ; Garrote, Estibaliz et al.
Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings. ed. / Jaime S. Cardoso; Roberto Paredes; Xosé M. Pardo. Springer Verlag, 2015. p. 587-594 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9117).

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

TY - GEN

T1 - Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition

AU - Galdran, Adrian

AU - Picón, Artzai

AU - Garrote, Estibaliz

AU - Pardo, David

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2015.

PY - 2015

Y1 - 2015

N2 - Pectoral muscle segmentation on medio-lateral oblique views of mammograms represents an important preprocessing step in many mammographic image analysis tasks. Although its location can be perceptually obvious for a human observer, the variability in shape, size, and intensities of the pectoral muscle boundary turns its automatic segmentation into a challenging problem. In this work we propose to decompose the input mammogram into its textural and structural components at different scales prior to dynamically thresholding it into several levels. The resulting segmentations are refined with an active contour model and merged together by means of a simple voting scheme to remove possible outliers. Our method performs well compared to several other state-ofthe- art techniques. An average DICE similarity coefficient of 0.91 and mean Hausdorff distance of 3.66 ± 3.23 mm. validate our approach.

AB - Pectoral muscle segmentation on medio-lateral oblique views of mammograms represents an important preprocessing step in many mammographic image analysis tasks. Although its location can be perceptually obvious for a human observer, the variability in shape, size, and intensities of the pectoral muscle boundary turns its automatic segmentation into a challenging problem. In this work we propose to decompose the input mammogram into its textural and structural components at different scales prior to dynamically thresholding it into several levels. The resulting segmentations are refined with an active contour model and merged together by means of a simple voting scheme to remove possible outliers. Our method performs well compared to several other state-ofthe- art techniques. An average DICE similarity coefficient of 0.91 and mean Hausdorff distance of 3.66 ± 3.23 mm. validate our approach.

KW - Breast cancer detection

KW - Computer-aided diagnosis

KW - Mammographic image analysis

KW - Pectoral muscle segmentation

UR - https://www.scopus.com/pages/publications/84937509876

U2 - 10.1007/978-3-319-19390-8_66

DO - 10.1007/978-3-319-19390-8_66

M3 - Conference contribution

AN - SCOPUS:84937509876

SN - 978-3-319-19390-8

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

SP - 587

EP - 594

BT - Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings

A2 - Cardoso, Jaime S.

A2 - Paredes, Roberto

A2 - Pardo, Xosé M.

PB - Springer Verlag

T2 - 7th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2015

Y2 - 17 June 2015 through 19 June 2015

ER -

Galdran A , Picón A , Garrote E, Pardo D. Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition. In Cardoso JS, Paredes R, Pardo XM, editors, Pattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings. Springer Verlag. 2015. p. 587-594. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-19390-8_66

Pectoral muscle segmentation in mammograms based on cartoon-texture decomposition

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this