Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images

Hanzheng Wang, Randy H. Moss, Xiaohe Chen, R. Joe Stanley, William V. Stoecker, M. Emre Celebi, Joseph M. Malters, James M Grichnik, Ashfaq A. Marghoob, Harold S. Rabinovitz, Scott W. Menzies, Thomas M. Szalapski

Research output: Contribution to journalArticle

35 Citations (Scopus)

Abstract

In previous research, a watershed-based algorithm was shown to be useful for automatic lesion segmentation in dermoscopy images, and was tested on a set of 100 benign and malignant melanoma images with the average of three sets of dermatologist-drawn borders used as the ground truth, resulting in an overall error of 15.98%. In this study, to reduce the border detection errors, a neural network classifier was utilized to improve the first-pass watershed segmentation; a novel "edge object value (EOV) threshold" method was used to remove large light blobs near the lesion boundary; and a noise removal procedure was applied to reduce the peninsula-shaped false-positive areas. As a result, an overall error of 11.09% was achieved.

Original languageEnglish
Pages (from-to)116-120
Number of pages5
JournalComputerized Medical Imaging and Graphics
Volume35
Issue number2
DOIs
StatePublished - Mar 1 2011

Fingerprint

Post and Core Technique
Dermoscopy
Watersheds
Skin
Error detection
Processing
Noise
Melanoma
Classifiers
Neural networks
Light
Research

Keywords

  • Image processing
  • Malignant melanoma
  • Neural network
  • Segmentation
  • Watershed

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Health Informatics
  • Radiological and Ultrasound Technology
  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition

Cite this

Wang, H., Moss, R. H., Chen, X., Stanley, R. J., Stoecker, W. V., Celebi, M. E., ... Szalapski, T. M. (2011). Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images. Computerized Medical Imaging and Graphics, 35(2), 116-120. https://doi.org/10.1016/j.compmedimag.2010.09.006

Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images. / Wang, Hanzheng; Moss, Randy H.; Chen, Xiaohe; Stanley, R. Joe; Stoecker, William V.; Celebi, M. Emre; Malters, Joseph M.; Grichnik, James M; Marghoob, Ashfaq A.; Rabinovitz, Harold S.; Menzies, Scott W.; Szalapski, Thomas M.

In: Computerized Medical Imaging and Graphics, Vol. 35, No. 2, 01.03.2011, p. 116-120.

Research output: Contribution to journalArticle

Wang, H, Moss, RH, Chen, X, Stanley, RJ, Stoecker, WV, Celebi, ME, Malters, JM, Grichnik, JM, Marghoob, AA, Rabinovitz, HS, Menzies, SW & Szalapski, TM 2011, 'Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images', Computerized Medical Imaging and Graphics, vol. 35, no. 2, pp. 116-120. https://doi.org/10.1016/j.compmedimag.2010.09.006
Wang, Hanzheng ; Moss, Randy H. ; Chen, Xiaohe ; Stanley, R. Joe ; Stoecker, William V. ; Celebi, M. Emre ; Malters, Joseph M. ; Grichnik, James M ; Marghoob, Ashfaq A. ; Rabinovitz, Harold S. ; Menzies, Scott W. ; Szalapski, Thomas M. / Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images. In: Computerized Medical Imaging and Graphics. 2011 ; Vol. 35, No. 2. pp. 116-120.
@article{4589324e43934817bd39e31b6a090896,
title = "Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images",
abstract = "In previous research, a watershed-based algorithm was shown to be useful for automatic lesion segmentation in dermoscopy images, and was tested on a set of 100 benign and malignant melanoma images with the average of three sets of dermatologist-drawn borders used as the ground truth, resulting in an overall error of 15.98{\%}. In this study, to reduce the border detection errors, a neural network classifier was utilized to improve the first-pass watershed segmentation; a novel {"}edge object value (EOV) threshold{"} method was used to remove large light blobs near the lesion boundary; and a noise removal procedure was applied to reduce the peninsula-shaped false-positive areas. As a result, an overall error of 11.09{\%} was achieved.",
keywords = "Image processing, Malignant melanoma, Neural network, Segmentation, Watershed",
author = "Hanzheng Wang and Moss, {Randy H.} and Xiaohe Chen and Stanley, {R. Joe} and Stoecker, {William V.} and Celebi, {M. Emre} and Malters, {Joseph M.} and Grichnik, {James M} and Marghoob, {Ashfaq A.} and Rabinovitz, {Harold S.} and Menzies, {Scott W.} and Szalapski, {Thomas M.}",
year = "2011",
month = "3",
day = "1",
doi = "10.1016/j.compmedimag.2010.09.006",
language = "English",
volume = "35",
pages = "116--120",
journal = "Computerized Medical Imaging and Graphics",
issn = "0895-6111",
publisher = "Elsevier Limited",
number = "2",

}

TY - JOUR

T1 - Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images

AU - Wang, Hanzheng

AU - Moss, Randy H.

AU - Chen, Xiaohe

AU - Stanley, R. Joe

AU - Stoecker, William V.

AU - Celebi, M. Emre

AU - Malters, Joseph M.

AU - Grichnik, James M

AU - Marghoob, Ashfaq A.

AU - Rabinovitz, Harold S.

AU - Menzies, Scott W.

AU - Szalapski, Thomas M.

PY - 2011/3/1

Y1 - 2011/3/1

N2 - In previous research, a watershed-based algorithm was shown to be useful for automatic lesion segmentation in dermoscopy images, and was tested on a set of 100 benign and malignant melanoma images with the average of three sets of dermatologist-drawn borders used as the ground truth, resulting in an overall error of 15.98%. In this study, to reduce the border detection errors, a neural network classifier was utilized to improve the first-pass watershed segmentation; a novel "edge object value (EOV) threshold" method was used to remove large light blobs near the lesion boundary; and a noise removal procedure was applied to reduce the peninsula-shaped false-positive areas. As a result, an overall error of 11.09% was achieved.

AB - In previous research, a watershed-based algorithm was shown to be useful for automatic lesion segmentation in dermoscopy images, and was tested on a set of 100 benign and malignant melanoma images with the average of three sets of dermatologist-drawn borders used as the ground truth, resulting in an overall error of 15.98%. In this study, to reduce the border detection errors, a neural network classifier was utilized to improve the first-pass watershed segmentation; a novel "edge object value (EOV) threshold" method was used to remove large light blobs near the lesion boundary; and a noise removal procedure was applied to reduce the peninsula-shaped false-positive areas. As a result, an overall error of 11.09% was achieved.

KW - Image processing

KW - Malignant melanoma

KW - Neural network

KW - Segmentation

KW - Watershed

UR - http://www.scopus.com/inward/record.url?scp=79951515493&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79951515493&partnerID=8YFLogxK

U2 - 10.1016/j.compmedimag.2010.09.006

DO - 10.1016/j.compmedimag.2010.09.006

M3 - Article

C2 - 20970307

AN - SCOPUS:79951515493

VL - 35

SP - 116

EP - 120

JO - Computerized Medical Imaging and Graphics

JF - Computerized Medical Imaging and Graphics

SN - 0895-6111

IS - 2

ER -