Classification of Breast Mass in Mammography with an Improved Level Set Segmentation by Combining Morphological Features and Texture Features

Abstract

In this chapter, an improved level set-based mass segmentation is proposed. The new mass segmentation algorithm is based on a new energy functional, which combines energy terms from several previous algorithms. Experimental results show that the proposed level set segmentation algorithm can get better segmentation results than the original one. Based on the contours obtained by the improved level set algorithm, mass regions are classified into malignant or benign using linear discriminant analysis (LDA) and support vector machine (SVM) by combining morphological features and texture features. Experiments show that the accuracy of mass classification by combining morphological features and textures features is higher than the accuracy of mass classification using only morphological features or texture features.