A Comparative Study of Fuzzy Thresholding Techniques for Mass Detection in Digital Mammography

Title
A Comparative Study of Fuzzy Thresholding Techniques for Mass Detection in Digital Mammography
Publication Date
2012
Author(s)
Alharbi, Hajar Mohammedsaleh H
Kwan, Paul H
Sajeev, Abudulkadir
Editor
Editor(s): Brendan McCane, Steven Mills, Jeremiah D Deng
Type of document
Conference Publication
Language
en
Entity Type
Publication
Publisher
Association for Computing Machinery (ACM)
Place of publication
New York, United States of America
Series
International Conference Proceedings Series (ICPS)
UNE publication id
une:12027
Abstract
Segmenting suspicious regions in mammographic images that may contain tumours from the background parenchyma of the breast is a highly challenging task. This is made difficult by factors including the complicated structure of breast tissues, unclear boundaries between normal tissues and tumours, and the low contrast between masses and surrounding regions in the images. In recent years, many researchers have discovered that fuzzy-logic based techniques have a number of advantages over conventional crisp approaches in segmenting masses in mammographic images. To this end, we compare five representative fuzzy thresholding techniques for this task in this paper using the recall and precision metrics. Experimental results revealed that fuzzy similarity thresholding achieves higher segmentation accuracy over a test set of 54 mammographic images selected from the mini-MIAS database.
Link
Citation
IVCNZ '12: Proceedings of the 27th International Conference on Image and Vision Computing New Zealand, p. 330-334
ISBN
9781450314732
Start page
330
End page
334

Files:

NameSizeformatDescriptionLink