Multiscale tumor detection and segmentation in mammograms

Breast cancer is a leading cause of death among women. Automatic computer-aided analysis method can help radiologists to make more reliable and efficient decisions. Tumors and other abnormalities are of basic interests that need to be segmented and extracted in mammograms. In this paper, a new wavelet-packet multiscale image segmentation method is presented first. By combining this new method with a multiscale region-based segmentation method, we develop a new generic systematic scheme to segment suspicious tumor areas with exact boundaries in mammograms based on multiscale analysis in both grayscale and space. Examples are presented to illustrate the effectiveness of the technique. The new segmentation scheme can also be used in analysis of other abnormal tissues in mammograms.