Microcalcification segmentation and mammogram image enhancement using nonlinear filtering

Microcalci cations are tiny calcium deposits in breast tissues. They appear as small bright spots on mammograms. Since microcalci cations are small and subtle abnormalities, they may be overlooked by an examining radiologist. In this work, we propose microcalci cation segmentation and mammogram enhancement methods based on nonlinear subband decomposition structures and outlier detection. The mammogram image is rst processed by a nonlinear lter bank, then the locations of the outliers are determined by outlier labeling method. The proposed method can also be used in the three-dimensional reconstruction of microcalci cation clusters. Simulation studies are presented.