NT & RP Journal
VISUAL ENHANCEMENT OF MICROCALCIFICATIONS AND MASSES IN DIGITAL MAMMOGRAMS USING MODIFIED MULTIFRACTAL ANALYSIS
......: info......: history......: editorial......: archive......: for authors......: subscription

Vol. XXX, No. 1, Pp. 01-82
March 2015
UDC 621.039+614.876:504.06
ISSN 1451-3994

Back to Contents

 


Pages: 61-69

Authors:
Tomislav M. Stojiæ

Abstract


Microcalcifications and masses, as breast tissue anomalies (deviations from observed background regularity), may be viewed as statistically rare occurrences in a mammogram image. After recognizing their principal common features – bright image parts not belonging to the surrounding tissue, with significant local contrast just around the edges – several modifications to multifractal image analysis have been introduced. Starting from a mammogram image, the proposed method creates corresponding multifractal images. Additional post-processing, based on mathematical morphology, refines the procedure by selecting and outlining only regions with possible microcalcifications and masses. The proposed method was tested through referent mammograms from the MiniMIAS database. In all cases involving the said database, the method has successfully enhanced declared anomalies: microcalcifications and masses. The results obtained have shown that the described procedure may provide visual assistance to radiologists in clinical mammogram examinations or be used as a preprocessing step for further mammogram processing, such as segmentation, classification, and automatic detection of suspected bright breast tissue lesions.

Key words: mammography, multifractal analysis, microcalcifications, masse, image processing

FULL PAPER IN PDF FORMAT (2.58 MB)

Vinča Institute of Nuclear Sciences :: Designed by milas :: July 2007
Operated by acapanic :: Last updated on May, 2015