An automated image analysis tool is being developed for the estimation of m
ammographic breast density. This tool may be useful for risk estimation or
for monitoring breast density change in prevention or intervention programs
. In this preliminary study, a data set of 4-view mammograms from 65 patien
ts was used to evaluate our approach. Breast density analysis was performed
on the digitized mammograms in three stages. First, the breast region was
segmented from the surrounding background by an automated breast boundary-t
racking algorithm. Second, an adaptive dynamic range compression technique
was applied to the breast image to reduce the range of the gray level distr
ibution in the low frequency background and to enhance the differences in t
he characteristic features of the gray level histogram for breasts of diffe
rent densities. Third, rule-based classification was used to classify the b
reast images into four classes according to the characteristic features of
their gray level histogram. For each image, a gray level threshold was auto
matically determined to segment the dense tissue from the breast region. Th
e area of segmented dense tissue as a percentage of the breast area was the
n estimated. To evaluate the performance of the algorithm, the computer seg
mentation results were compared to manual segmentation with interactive thr
esholding by five radiologists. A "true" percent dense area for each mammog
ram was obtained by averaging the manually segmented areas of the radiologi
sts. We found that the histograms of 6% (8 CC and 8 MLO views) of the breas
t regions were misclassified by the computer, resulting in Door segmentatio
n of the dense region. For the images with correct classification, the corr
elation between the computer-estimated percent dense area and the ''truth''
was 0.94 and 0.91, respectively, for CC and MLO views, with a mean bias of
less than 2%. The mean biases of the five radiologists' visual estimates f
or the same images ranged from 0.1% to 11%. The results demonstrate the fea
sibility of estimating mammographic breast density using computer vision te
chniques and its potential to improve the accuracy and reproducibility of b
reast density estimation in comparison with the subjective visual assessmen
t by radiologists. (C) 2001 American Association of Physicists in Medicine.