Early detection of breast cancer using mathematical morphology

Abstract

In this paper a method for detection of microcalcifications (MCCs), which appear at the beginning stages of breast cancer, is presented by using mathematical morphology. This method consists of four steps. At the first step the contrast of suspicious regions on the image is enhanced by contrast stretching. In the next step a morphological filtering process is applied by using six different structured elements (SEs). At the same step substruction image is obtained by substructing the filtered image from the original image. In the next step a black and white image is generated from this substraction image by applying a threshold, which is calculated by using histogam. Finally, healthy regions are classified by refiltering black and white image using morphological opening. The proposed method was tested on 17 mammograms and 82.2% success rate was observed.