A novel method of extracting and classifying the features of masses in mammograms

Abstract

Some improvements in the classification of masses in the breast are proposed in this paper. First, for the purpose of enriching the information concerning the shape of the mass, a new morphological feature is extracted. Then, the textural features of the region of interest (ROI) are extracted by combining the undecimated wavelet transform (UWT) and the gray level co-occurrence matrix (GLCM). Finally, based on the geometrical and textural features, the feature-weighted support-vector machine (FWSVM) is used to distinguish between malignant and benign masses. The experiments implemented on the public digital database for screening mammography (DDSM) indicated that the proposed improvements can achieve better results than the existing methods.