A controlled phantom study of a noise equalization algorithm for detecting microcalcifications in digital mammograms

We report on some extensions and further developments of a well-known microcalcification detection algorithm based on adaptive noise equalization. Tissue equivalent phantom images with and without labeled microcalcifications were subjected to this algorithm, and analyses of results revealed some sho...

Full description

Saved in:

Bibliographic Details
Published in:	Medical physics (Lancaster) Vol. 28; no. 4; pp. 445 - 454
Main Authors:	Gürün, Özgür Ozan, Fatouros, Panos P., Kuhn, Gary M., Shaw de Paredes, Ellen
Format:	Journal Article
Language:	English
Published:	United States American Association of Physicists in Medicine 01-04-2001
Subjects:	Algorithms Breast - pathology digital mammography Digital radiography Female Humans Image analysis Image quality image transforms local contrast Mammography Mammography - instrumentation Mammography - methods Medical image contrast medical image processing Medical imaging microcalcifications Models, Statistical Models, Theoretical noise noise equalization Phantoms, Imaging Physicists Statistical properties image transforms noise equalization local contrast digital mammography microcalcifications
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	We report on some extensions and further developments of a well-known microcalcification detection algorithm based on adaptive noise equalization. Tissue equivalent phantom images with and without labeled microcalcifications were subjected to this algorithm, and analyses of results revealed some shortcomings in the approach. Particularly, it was observed that the method of estimating the width of distributions in the feature space was based on assumptions which resulted in the loss of similarity preservation characteristics. A modification involving a change of estimator statistic was made, and the modified approach was tested on the same phantom images. Other modifications for improving detectability such as downsampling and use of alternate local contrast filters were also tested. The results indicate that these modifications yield improvements in detectability, while extending the generality of the approach. Extensions to real mammograms and further directions of research are discussed.
Bibliography:	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23
ISSN:	0094-2405 2473-4209
DOI:	10.1118/1.1359245