The segmentation of a digital mammogram into the breast region and the back
ground is a necessary prerequisite in computer-assisted diagnosis of mammog
rams. By the exclusion of the background region, the accuracy of the analys
is is increased and the running-time is decreased. The algorithm which segm
ents the breast region from the background should be fully automated and gi
ve correct results for all kinds of digitized mammograms, including low-qua
lity images. In this paper we present such an algorithm based on histogram
thresholding, morphological filtering and contour modeling. Quantitative te
st results indicate that the computed boundary follows the estimated bounda
ry accurately. (C) 2000 Elsevier Science Ltd. All rights reserved.