Feature Extraction from contours shape for tumor analyzing in Mammographic images

The cancer treatment is effective only, if it is detected at an early stage. In this context, Mammography is the most efficient method for early detection. Due to the complexity of this last, the distinction of microcalcifications or opacities is very difficult. This paper deals with the problem of shape feature extraction in digital mammograms, particularly, the boundary information. In fact, we evaluated the efficiency on boundary information possessed by mass region. We propose a feature vector based on boundary analysis to ameliorating three caracteristics like RDM, convexity and angular ones. We use the Digital Database for Screening Mammography DDSM for experiments. Sonic classifiers like Multilayer Perceptron MLP and k-Nearest Neighbors kNN are used to distinguish the pathological records from the healthy ones. Using MLP classifiers we obtained 94,2% as sensitivity( percentage of pathological ROIs correctly classified). The results in term of specificity (percentage of non-pathological ROIs correctly classified) grows around 97,9% using MLP classifier