Comparative study of features classification methods for mass lesion recognition in digitized mammograms