News |  Sitemap |  Contact
PDF Export  | 

Detection and dissection of the CDMAM phantom test image in mammography for a blind and randomised comparative visual test



A plugin for ImageJ was developed to detect and to dissect the CDMAM (Contrast Detail Mammography) phantom image used in mammography. The phantom, containing multiple square regions with gold discs of different thickness and diameter, is used in visual constancy test methods regarding european guidelines. A comprehensive study has shown, that this test method is subject to a profound inherent error, due to a human perception bias. This program dissects the phantom image to a randomised image stack with all square regions, which is then evaluated visually without the specified disadvantage or could be evaluated automatically.

Material and Methods:

The phantom test image contains a predominant pattern, a set of lines, which divide the image into square sections, containing the gold discs. The Hough Transform is used to detect those lines, since it is a suitable mathematical method for line detection. Nonetheless some preceding processing, like noise reduction and equalization of vignetting, is necessary for a proper discrimination of the peaks in the Hough domain. Discrete derivation and the correlation coefficient are now successful means for this task, since the line pattern is highly regular. Low-pass filtering in the coreesponding area of the Hough transform is the last step for the detection of the lines. Finally an ImageJ image stack is created from all dissected areas and presented for the subsequent visual or automatic test.


A comprehensive study has shown, that a randomised presentation of the test areas of the CDMAM phantom on the monitor is indispensable for an unbiased and reliable detection by examiners. This ImageJ plugin creates rapidly an image stack from a CDMAM phantom image, containing all test regions with gold discs.

Digital mammography, CDMAM (Contrast Detail Mammography), medical imaging, acceptance test, constancy test, performance test

Christian Blendl

Cologne University of Applied Science, Institute of Media and Imaging Technology


Short Biography   

Diploma in Photo Engineering and Media Technology acquired at the Cologne University of Applied Science, Institute of Media and Imaging Technology. Momentary doing a master thesis at the same faculty in the Masters course Media and Imaging Technology.

Special interest and knowledge: digital image processing in medical images, Java programming and ImageJ, developing algorithms for the solution of special problems in digital mammography.

© Luxembourg Institute of Science and Technology | Legal Notice