Computer aided mass segmentation in mammogram images using Grey wolf Optimized Region growing technique

One of the dangerous threats, that affect women all around the globe is breast cancer, leading to early mortality in women. According to researchers the survival rate of the breast cancer affected person can be improved by a greater amount by its early detection. Hence, there is need for development of an automated system, which can act as an aid for supporting the radiologists in making proper diagnostic decision. The proposed work involves detection of the breast masses by making use of an optimized region growing method, in which the optimal seed point selection and optimal threshold generation was achieved using Grey Wolf Optimization (GWO). In the proposed work the extraction of both global and local features are being considered. The global features considered includes shape features, Grey Level Co-occurrence Matrix (GLCM) and Grey Level Run Length Matrix (GLRLM) for extracting texture feature and local texture feature is extracted using Local Binary Pattern (LBP) and Scale invariant feature transform (SIFT). The fusion of the local and global features were being fed to Support Vector Machine (SVM) classifier, which differentiates the masses as either benign or malignant in nature. The proposed methodology achieved a highest accuracy of 96% by the fusion of global texture feature GLCM and LBP.