Effective detection of mass abnormalities and its classification using multi-SVM classifier with digital mammogram images

Abstract

Breast cancer is one of the most common kind of cancer, as well as it's the major cause in increasing mortality rate in women. Mammography is the effective method that is used for the early detection of breast cancer. Digital mammograms have become the most effective source for the detection of breast cancer. This paper proposes a method for the detection and classification of mass abnormalities in digital mammogram images using multi SVM classifier. The goal of this research is to increase the diagnostic accuracy of image processing and optimum classification between malignant and benign abnormalities in mass region which reduces the misclassification of breast images. Malignant and benign abnormalities are detected from the segmented images using region based segmentation, which correspond to the Regions of Interest (ROIs) or abnormal regions. Texture based features are extracted from the ROI samples using Gray Level Co-Occurrence Matrices (GLCMs). For the purpose of classification between malignant and benign samples, the optimum subset of texture features are classified using a Multi-Support Vector Machine (SVM). The effectiveness of this paper is examined using classification accuracy, which produced an accuracy of 94%.