Feature extraction and support vector machine based classification for false positive reduction in mammographic images

Abstract

In this paper, we propose a new method for massive false positive reduction in. Our goal is to distinguish between the true recognized masses and the ones which actually normal parenchyma. Our proposal is based on Block Difference Inverse Probability (BDIP) and Support Vector Machine (SVM) for classifying the detected masses. The proposed approach is evaluated in about 2700 ROIs detected from Mini-MIAS database. An accuracy of Az = 0.91 (area under the curve) is obtained. © Springer Science+Business Media Dordrecht 2014.