The purpose of this study was to develop and evaluate a computerized m
ethod of calculating a breast density index (BDI) from digitized mammo
grams that was designed specifically to model radiologists' perception
of breast density. A set of 153 pairs of digitized mammograms (cranio
-caudal, CC, and mediolateral oblique, MLO, views) were acquired and p
reprocessed to reduce detector biases. The sets of mammograms were ord
ered on an ordinal scale (a scale based only on relative rank-ordering
) by two radiologists, and a cardinal (an absolute numerical score) BD
I value was calculated from the ordinal ranks. The images were also as
signed cardinal BDI values by the radiologists in a subsequent session
. Six mathematical features (including fractal dimension and others) w
ere calculated from the digital mammograms, and were used in conjuncti
on with single value decomposition and multiple linear regression to c
alculate a computerized BDI. The linear correlation coefficient betwee
n different ordinal ranking sessions were as follows: intraradiologist
intraprojection (CC/CC): r = 0.978; intraradiologist interprojection
(CC/MLO): r = 0.960; and interradiologist intraprojection (CC/CC): r =
0.968. A separate breast density index was derived from three separat
e ordinal rankings by one radiologist (two with CC views, one with the
MLO view). The computer derived BDI had a correlation coefficient (r)
of 0.907 with the radiologists' ordinal BDI. A comparison between rad
iologists using a cardinal scoring system (which is closest to how rad
iologists actually evaluate breast density) showed r = 0.914. A breast
density index calculated by a computer but modeled after radiologist
perception of breast density may be valuable in objectively measuring
breast density. Such a metric may prove valuable in numerous areas, in
cluding breast cancer risk assessment and in evaluating screening tech
niques specifically designed to improve imaging of the dense breast. C
opyright (C) 1998 by W.B. Saunders Company.