Digital mammogram images can have different kinds of artifacts that affect the accuracy of the detection of tumor tissues in the automated computer-aided detection (CAD) system for mammograms. Preprocessing to remove such artifacts is an important step. In this paper, a preprocessing technique for digital mammograms is devised which removes labels, scanning artifacts and the pectoral muscle. First, it removes hurdles like labels, scanning and taping artifacts using an automated algorithm based on thresholding. Then, using the active contours and the proposed stopping algorithm it obtains the contour which contains the boundary of the pectoral muscle. Later, it extracts the pectoral muscle binary image from the contour. Finally, using the pectoral muscle binary image and the original mammogram image it obtains the desired image without any artifacts and the pectoral muscle. We tested the proposed algorithm on the mammograms from the mini-MIAS database and it worked very efficiently. It provided very effective and accurate results for pectoral muscle segmentation. It provided up to 97.84% accuracy, computed from well segmented results.


Active contours, Computer-aided detection, Mammogram preprocessing, Pectoral muscle segmentation, Region of interest.