Breast cancer is the most common and deadly cancer among women and early diagnosis is essential to reduce the mortality rate. Histopathological exams are invasive and usually performed to confirm the malignancy of lesions found in exams. To differentiate between categories 3 and 4 of the BI-RADS lexicon, patients often undergo histopathological examinations, and, in most cases, the lesions are benign. For this detection to be more accurate and for patients to be less subjected to clinical histopathology procedures, CAD systems (computer-aided diagnostic systems) have been developed to complement the diagnosis and assist the radiologist in making decisions in case of doubts. CAD systems work through image processing, lesion segmentation, selection of quantitative parameters and classification. The automatic segmentation of breast lesions can be less accurate because there is energy loss during propagation and in the formation of the image, especially in the lower part of the lesion. Thus, the aim of this study is to verify if there is a significant difference in the classification of lesions when only the upper half of the segmentation is used, compared to the complete segmentation of the lesion.