Novel technique for the detection of abnormalities in Mammograms using texture and geometric features

The paper investigates on recognition of breast abnormalities. A novel feature frame work was proposed on mammographic patches based on both texture and geometric features for classification of breast tissues into normal, malignant and benign. The methodology comprises of five stages. First step is preprocessing, texture feature extraction using Local quinary pattern for classifying breast tissues into normal and abnormal, Automatic segmentation of mass using k means algorithm, a new geometric feature descriptors extraction to classify them into benign and malignant and two stage classification. Our feature extraction method attained 99.27 for normal and abnormal, 79.41% for benign and malignant and over all accuracy for three class classification is 89.05%.