Breast density estimation: Correlation of mammographic density and MR volumetric density

Studies have demonstrated a strong correlation between mammographic breast density and breast cancer risk. Mammographic breast density may therefore be used as a surrogate marker for monitoring the response to treatment in studies of breast cancer prevention or intervention methods. In this study, we evaluated the accuracy of using mammograms for estimating breast density by analyzing the correlation between the percent mammographic dense area and the percent glandular tissue volume as estimated from MR images. A data set of 37 patients who had corresponding MR images and mammograms was collected. The glandular tissue regions in the MR slices were segmented by a semi-automatic method and the percent glandular tissue volume calculated. Mammographic breast density was estimated by an automated image analysis program. It was found that the correlation between the percent dense area of the CC and MLO views and the percent volumetric fibroglandular tissue on MR images was 0.93 and 0.91, respectively, with a mean bias of 4.4%. The high correlation indicates the usefulness of mammographic density as a surrogate for breast density estimation.