Fully automated breast boundary and pectoral muscle segmentation in mammograms

Andrik Rampun, PJ Morrow, BW Scotney, RJ Winder

Research output: Contribution to journalArticle

26 Citations (Scopus)

Abstract

Breast and pectoral muscle segmentation is an essential pre-processing step for the subsequent processes in Computer Aided Diagnosis (CAD) systems. Esti- mating the breast and pectoral boundaries is a difficult task especially in mam- mograms due to artifacts, homogeneity between the pectoral and breast regions, and low contrast along the skin-air boundary. In this paper, a breast bound- ary and pectoral muscle segmentation method in mammograms is proposed. For breast boundary estimation, we determine the initial breast boundary via thresholding and employ Active Contour Models without edges to search for the actual boundary. A post-processing technique is proposed to correct the overestimated boundary caused by artifacts. The pectoral muscle boundary is estimated using Canny edge detection and a pre-processing technique is pro- posed to remove noisy edges. Subsequently, we identify five edge features to find the edge that has the highest probability of being the initial pectoral con- tour and search for the actual boundary via contour growing. The segmentation results for the proposed method are compared with manual segmentations using322, 208 and 100 mammograms from the Mammographic Image Analysis Soci- ety (MIAS), INBreast and Breast Cancer Digital Repository (BCDR) databases, respectively. Experimental results show that the breast boundary and pectoral muscle estimation methods achieved dice similarity coefficients of 98.8% and 97.8% (MIAS), 98.9% and 89.6% (INBreast) and 99.2% and 91.9% (BCDR), respectively.
LanguageEnglish
Pages28-41
JournalArtificial Intelligence in Medicine
Volume79
Early online date9 Jun 2017
DOIs
Publication statusE-pub ahead of print - 9 Jun 2017

Fingerprint

Pectoralis Muscles
Muscle
Breast
Image analysis
Processing
Computer aided diagnosis
Edge detection
Artifacts
Post and Core Technique
Skin
Breast Neoplasms
Air
Databases

Keywords

  • Breast mammography
  • Breast segmentation
  • Pectoral segmentation
  • Computer aided diagnosis

Cite this

@article{9d8a5b41efc148689d741ece683889ec,
title = "Fully automated breast boundary and pectoral muscle segmentation in mammograms",
abstract = "Breast and pectoral muscle segmentation is an essential pre-processing step for the subsequent processes in Computer Aided Diagnosis (CAD) systems. Esti- mating the breast and pectoral boundaries is a difficult task especially in mam- mograms due to artifacts, homogeneity between the pectoral and breast regions, and low contrast along the skin-air boundary. In this paper, a breast bound- ary and pectoral muscle segmentation method in mammograms is proposed. For breast boundary estimation, we determine the initial breast boundary via thresholding and employ Active Contour Models without edges to search for the actual boundary. A post-processing technique is proposed to correct the overestimated boundary caused by artifacts. The pectoral muscle boundary is estimated using Canny edge detection and a pre-processing technique is pro- posed to remove noisy edges. Subsequently, we identify five edge features to find the edge that has the highest probability of being the initial pectoral con- tour and search for the actual boundary via contour growing. The segmentation results for the proposed method are compared with manual segmentations using322, 208 and 100 mammograms from the Mammographic Image Analysis Soci- ety (MIAS), INBreast and Breast Cancer Digital Repository (BCDR) databases, respectively. Experimental results show that the breast boundary and pectoral muscle estimation methods achieved dice similarity coefficients of 98.8{\%} and 97.8{\%} (MIAS), 98.9{\%} and 89.6{\%} (INBreast) and 99.2{\%} and 91.9{\%} (BCDR), respectively.",
keywords = "Breast mammography, Breast segmentation, Pectoral segmentation, Computer aided diagnosis",
author = "Andrik Rampun and PJ Morrow and BW Scotney and RJ Winder",
year = "2017",
month = "6",
day = "9",
doi = "10.1016/j.artmed.2017.06.001",
language = "English",
volume = "79",
pages = "28--41",
journal = "Artificial Intelligence in Medicine",
issn = "0933-3657",
publisher = "Elsevier",

}

Fully automated breast boundary and pectoral muscle segmentation in mammograms. / Rampun, Andrik; Morrow, PJ; Scotney, BW; Winder, RJ.

In: Artificial Intelligence in Medicine, Vol. 79, 09.06.2017, p. 28-41.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Fully automated breast boundary and pectoral muscle segmentation in mammograms

AU - Rampun, Andrik

AU - Morrow, PJ

AU - Scotney, BW

AU - Winder, RJ

PY - 2017/6/9

Y1 - 2017/6/9

N2 - Breast and pectoral muscle segmentation is an essential pre-processing step for the subsequent processes in Computer Aided Diagnosis (CAD) systems. Esti- mating the breast and pectoral boundaries is a difficult task especially in mam- mograms due to artifacts, homogeneity between the pectoral and breast regions, and low contrast along the skin-air boundary. In this paper, a breast bound- ary and pectoral muscle segmentation method in mammograms is proposed. For breast boundary estimation, we determine the initial breast boundary via thresholding and employ Active Contour Models without edges to search for the actual boundary. A post-processing technique is proposed to correct the overestimated boundary caused by artifacts. The pectoral muscle boundary is estimated using Canny edge detection and a pre-processing technique is pro- posed to remove noisy edges. Subsequently, we identify five edge features to find the edge that has the highest probability of being the initial pectoral con- tour and search for the actual boundary via contour growing. The segmentation results for the proposed method are compared with manual segmentations using322, 208 and 100 mammograms from the Mammographic Image Analysis Soci- ety (MIAS), INBreast and Breast Cancer Digital Repository (BCDR) databases, respectively. Experimental results show that the breast boundary and pectoral muscle estimation methods achieved dice similarity coefficients of 98.8% and 97.8% (MIAS), 98.9% and 89.6% (INBreast) and 99.2% and 91.9% (BCDR), respectively.

AB - Breast and pectoral muscle segmentation is an essential pre-processing step for the subsequent processes in Computer Aided Diagnosis (CAD) systems. Esti- mating the breast and pectoral boundaries is a difficult task especially in mam- mograms due to artifacts, homogeneity between the pectoral and breast regions, and low contrast along the skin-air boundary. In this paper, a breast bound- ary and pectoral muscle segmentation method in mammograms is proposed. For breast boundary estimation, we determine the initial breast boundary via thresholding and employ Active Contour Models without edges to search for the actual boundary. A post-processing technique is proposed to correct the overestimated boundary caused by artifacts. The pectoral muscle boundary is estimated using Canny edge detection and a pre-processing technique is pro- posed to remove noisy edges. Subsequently, we identify five edge features to find the edge that has the highest probability of being the initial pectoral con- tour and search for the actual boundary via contour growing. The segmentation results for the proposed method are compared with manual segmentations using322, 208 and 100 mammograms from the Mammographic Image Analysis Soci- ety (MIAS), INBreast and Breast Cancer Digital Repository (BCDR) databases, respectively. Experimental results show that the breast boundary and pectoral muscle estimation methods achieved dice similarity coefficients of 98.8% and 97.8% (MIAS), 98.9% and 89.6% (INBreast) and 99.2% and 91.9% (BCDR), respectively.

KW - Breast mammography

KW - Breast segmentation

KW - Pectoral segmentation

KW - Computer aided diagnosis

U2 - 10.1016/j.artmed.2017.06.001

DO - 10.1016/j.artmed.2017.06.001

M3 - Article

VL - 79

SP - 28

EP - 41

JO - Artificial Intelligence in Medicine

T2 - Artificial Intelligence in Medicine

JF - Artificial Intelligence in Medicine

SN - 0933-3657

ER -