Mammographic Mass Description for Breast Cancer Recognition

Due to the increasing number of mammograms in recent years, several techniques for automatic breast cancer recognition have been developed. These new methods have enabled the development of different Computer Aided Diagnosis systems often known by the acronym (CAD). The typical architecture of a CAD system is mainly composed of three major steps, features extraction, description and classification that leads to breast cancer recognition. In this chapter, and after presenting the features extraction approaches, we present geometric, morphologic and speculated mass descriptors. The comparison between new and very known descriptors, provides a considerable help on breast masses recognition. The recognition performance system is discussed through the obtained results using SVM classifier and ROC curves.