Document Type : Research Article
Digital mammography is most reliable and effective technique for early and accurate identification of Breast cancer. Image processing plays a significant role in diagnosis and classification of breast cancer in medical field. In this paper, a system is created to classify the mammogram images into three classes, namely Benign, Malignant and Normal. Mammogram images are pre-processed and the features are extracted from the segmented region. These features are used to train modified SVM and KNN classifier. The proposed Hybrid algorithm with modified SVM and KNN classifier helps to classify the mammogram images. This latest technique improves the SVM algorithm with introducing multi class for classification of breast cancer. It exploits the KNN algorithm according to the distribution of test images in a feature space. This study also evaluates the accuracy with the SVM and KNN classifier. The modified SVM and KNN hybrid algorithm produces higher prognosis accuracy than the KNN method and SVM technique. This method is tested for 10 test images with 20 trained. This methodology achieves an overall mean accuracy of 99.3406% in classification of mammogram images.
Keywords: Classification, KNN, MIAS, Proposed KNN with SVM.