Détection morphologique et texturale de pathologies mammographiques pour l’aide à la décision

Abstract

The segmentation of mammograms plays a major role in isolating areas, which can be subject to tumors. The identification of these zones is done in three stages: contrast enhancement, overdense regions detection and finally texture and shape analysis of the regions of interest for a classification of masses. We use region growing algorithm for the masses detection. Classification of these structures is accomplished through Support Vector Machines, which separate them into two groups, Benign Tumors and Malignant Tumors, using shape and texture descriptors to aid clinical diagnosis.