RSNA 2010 

Abstract Archives of the RSNA, 2010


An Automatic Framework for Assessing Breast Cancer Risk Due to Various Hormone Replacement Therapies (HRT): A Novel CAD Application in Digital Mammography

Education Exhibits

Presented on November 28, 2010


Gopal Karemore MS, Presenter: Nothing to Disclose
Sami Brandt PhD, Abstract Co-Author: Nothing to Disclose
Mads Nielsen PhD, Abstract Co-Author: Nothing to Disclose


It is well known that menopausal hormone therapy increases mammographic density. Increase in breast density may relate to breast cancer risk. Several computer assisted automatic methods for assessing mammographic density have been suggested by J.W. Byng (1996), N. Karssemeijer (1998), J.M. Boone(1998), S. Petroudi(2006), C. Tromans(2006) etc. All  these findings aims at reproducing the radiologist’s categorical rating system or at segmenting the dense tissue to get percentage density score. Since ninety percent of breast cancers arise from the ductal and lobular glands we choose to investigate features describing the local elongatedness or stripiness, especially trained to see the effect of HRT (Hormone Replacement Therapy ) thereby providing a non-subjective and reproducible measure and compare it to the BIRADS and percentage density measure.


Digitised mammographies of 2x135 completers of a two year, randomised, trial formed the base of the present analysis. Active treatments were transdermal estradiol releasing 0.014mg E2/week and orally administered raloxifene hydrochloride, 60mg/day respectively. Influence of the therapies on breast density was assessed with our computer-based heterogeneity examination of radiographs (E2-HER) measure which uses scale space features in order to examine the heterogeneity with in mammogram.


At baseline no mammography scoring methodology could separate the two treatment groups of transdermal estradiol and raloxifene. No treatment induced significant density changes measured by BI-RADS. Both treatments made the area percentage density increase and the estradiol significantly. Both treatments induced significant changes in E2-specific heterogeneity scoring (E2-HER) and the raloxifene treatment a significantly higher change.


The proposed automatic methodology shows substantial merit than radiologist scoring ie BIRADS and percentage density. This approach can be trained to detect changes due to HRT. The current study does not yield evidence against the hypothesis that “neither raloxifene nor low dose transdermal estradiol treatment increase the breast cancer risk”.


Cite This Abstract

Karemore, G, Brandt, S, Nielsen, M, An Automatic Framework for Assessing Breast Cancer Risk Due to Various Hormone Replacement Therapies (HRT): A Novel CAD Application in Digital Mammography.  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL.