Orientador: Prof. Dr. Marcelo Zanchetta do Nascimento / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia da Informação, 2014. / In this project is presented a new method to build a computer aided diagnosis system (CAD) for the classication of mammographic lesions. It was investigated the operators Local Binary Pattern (LBP), Completed Local Binary Pattern (CLBP), Center Symmetric Binary Pattern (CS-BP) and Centralized Binary Pattern (CBP) in this study. The proposed method was based in the application of Wavelet Transform, the LBP operators and their approximations. After obtaining the informations, the attributes reducer analyzes of variance (ANOVA) was applied in order to reduce the amount of information, thus eliminating data that were irrelevant to the classication. To evaluate the proposed method, the 10-fold cross validation technique was applied using the classiers Support Vector Machine (SVM), Random Forest (RaF) and Radial Basis Function (RBF). It was used regions of interest of mammographic images obtained from dierent databases: Digital Database for Screening Mammography (DDSM), Breast Cancer Digital Repository (BCDR) e INBreast. The results were demonstrated with the area under ROC curve (AUC), which were 1:0 for Benign versus Malignant and Benign versus Normal groups. The Normal versus Malignant group resulted in an AUC of 0:99 for the proposed method.