Mammogram Density Classification using Double Support Vector Machines
Yi‐Chong Zeng · 2018
Dense breast increases risk of breast cancer. There were of lesion diagnosis difficulties on dense breast for doctors, not least computer aided detection. In this paper, we propose a scheme to classify breast densities based on support vector machine (SVM) and voting system. First, 41 kinds of features are extracted from mammogram images. The first SVM is applied to the features for segmentation of mammary gland and fatty. Subsequently, one-dimension histogram is derived from score map of the first SVM. The second SVM is applied to the histogram to train multiple classifiers, and then classifies mammogram density. The experiment results demonstrate that the proposed scheme is capable of classifying breast density and is better than the compared approaches.