Texture analysis of gradient images for benign-malignant mass classification

Abstract

In this correspondence, texture analysis of gradient images has been analyzed for the categorization of mammo-graphic masses as benign or malignant. In addition to the local texture feature, Local Binary Pattern, approximation coefficients have been extracted from the gradient images using wavelet transform to evaluate their efficiency in a Computer-Aided Diagnosis (CADx) system. The experiments have been conducted with the DDSM database containing 200 mammograms where 10 fold cross validation technique has been incorporated with Fisher Linear Discriminant Analysis (FLDA) over the optimal set of features acquired via stepwise logistic regression method. A z value of 0.91 has been achieved as the best case which indicates an improvement over the results obtained with the normal mass region.