Classification of mammograms by breast density

Abstract

The risk of getting breast cancer is directly affected by the type of breast tissue predominant in the individual. The aim is to investigatehistogram-based image attributes in order to separate mammographic images by degree of breast density using the clustering technique. 75 mammographic images from the MIAS database were used, 25 of them belonging to each of the three classes: fatty, fatty-glandular and dense. After the selection of attributes, it obtained a 96% success rate in classifying the mammograms within the three classes when the attribute’s mean gray levels and the highest peak intensity of the histogram were used simultaneously in the clustering technique