Mammogram classification using back-propagation neural networks and texture feature descriptors

Abstract

Breast cancer has an important incidence in women worldwide. Early diagnosis of this illness plays a key role in decreasing its mortality and improves its prognosis. Currently, mammography is considered as the standard examination for detection of breast cancer. However, the identification of breast abnormalities and the classification of masses on mammographic images are not trivial tasks for dense breasts, and is a challenge for artificial intelligence and pattern recognition. This work presents preliminary results of automatic classification of mammographies by texture characterization based mainly on the Haralick's descriptors. We implement an artificial neural network (ANN) for classification in three classes: normal, benign and cancer using leave one out technique. The set of images for training and testing the ANN, are taken from the Digital Database for Screening Mammography (DDSM). Results show that the percentage of correct classification occurs in average for 84.72% of the data set.