Classification of Breast Cancer Histopathology Images using Texture Feature Analysis

In this paper, we propose a method for classification of histopathological images using texture features. The images are first segmented as epithelial lining surrounding the lumen for breast histopathology images using spatio-color-texture graph segmentation method. The features such as Gray Level Co-occurrence Matrix (GLCM), Graph Run Length Matrix (GRLM) features, and Euler number are extracted. The linear discriminant analyzer (LDA) is used to classify breast histology images. The performance of LDA classifier is compared with k-NN and SVM classifiers. The experiments and quantitative analysis shows that LDA classifier outperforms over others with 100% and 80% correct classification rate for the non-malignant Vs malignant breast histopathology images respectively.