An algorithm for detection and segmentation of clustered microcalcifications on mammograms

A reliable breast cancer diagnosis often depends upon an effective mammography screening program. In this paper, a three-stage procedure for enhancement, detection and segmentation of clustered microcalcifications is proposed. The first stage is to enhance the image, separate calcifications from the background tissue and normalize the whole image. In second stage, the spots of clustered microcalcifications are detected, from where we can determine the ROI. Then, in the third stage, clustered microcalcifications are segmented from the ROI. A great number of biopsy-proved mammograms are tested based on killed radiologists. The results show a promising potential of diagnosis on mammograms.