In this work, a comparative study of different segmentation techniques is carried out on INBREAST database, between, deep learning approach with the Convolutional Neural Networks and traditional machine learning techniques by the superpixel-based segmentation.