A Novel Classification Scheme for Breast Masses Based on Multi-View Information Fusion

The classification of breast masses into benign and malignant categories plays an important role in the area of computer-aided diagnosis (CAD) of breast cancer. In this paper, one novel scheme based on multi-view information fusion is proposed, in order to improve the accuracy and the robustness of the classification and reduce the false positive rates. Five contour and shape features of the masses are chosen, and two contour features are introduced. In this paper, 304 ROIs (regions of interest) that compose 152 multi-view pairs according to the corresponding breasts are chosen from DDSM dataset and divided into three groups, according to the characteristics of the masses in each pair. Different groups of ROIs are respectively used to check the effectiveness of the classification scheme and the experimental results of three ROIs groups demonstrate that the proposed classification scheme achieves a higher accuracy than those schemes using the individual classifiers in single-view, and the two introduced features could improve the classification performance significantly.