A new statistical-based approach for the discrimination of normal RoIs and RoIs depicting true masses is presented in this chapter. The method is based on modeling the tissue variation of both kinds of RoIs by extracting the principal components of a set of already classified RoIs. Subsequently, the system projects each new RoI onto a feature space that spans the significant variations among the known RoIs. The performance of the method is tested in two ways. Firstly, using ROC analysis, the method discriminates between both kinds of RoIs using a leave-one-out methodology. Secondly, the false positive reduction is integrated in the algorithm developed in the previous chapter in order to demonstrate its validity.