Classification of benign and malignant masses using bandelet and orthogonal ripplet type II transforms
Goriparthi Prathibha, Bollineni Sai Mohan · Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization · 2017
Breast cancer is one of the leading cause of death among women worldwide. Many CAD systems have been proposed for the early detection of cancer causing masses. In this paper, classification of mammograms from digital database for screening mammography (DDSM) is done using bandelet and orthogonal ripplet type II transforms. In this study, two subsets of mammograms are used from the DDSM database. The first subset contains 360 regions of interest (ROI) of mammograms obtained from howtek scanner and the second subset contains 300 ROIs of mammograms obtained from both lumisys and howtek scanner. Bandelet and orthogonal ripplet type II transform coefficients are extracted for these two subsets. First-order texture features are calculated for the ROIs using the coefficients of bandelet and orthogonal ripplet type II transform. Based on the first-order texture features, the ROIs are classified. The area under the curve for orthogonal ripplet type II transform is and bandelet transform is obtained using 360 mammograms of howtek scanner. Further the area under the curve for orthogonal ripplet type II transform is and bandelet transform is obtained using 300 mammograms of howtek and lumisys scanner for classifying benign and malignant masses. The values of obtained are superior to the existing wavelet-based transforms like wavelet, ridgelet, curvelet, contourlet and bandelet.