Multi resolution analysis for mass classification in digital mammogram using stochastic neighbor embedding

An efficient mass classification system for breast cancer in digital mammogram based on Discrete Wavelet Transform (DWT) and Stochastic Neighbor Embedding (SNE) technique is presented in this paper. The mass classification in digital mammogram is achieved by decomposing the mammogram image by using DWT at various levels. Then the high dimensional wavelet coefficients are reduced by using SNE. The reduced wavelet coefficients are used as features for the corresponding mammogram image to classify the given mammogram into normal, benign and malignant mass. In the proposed system KNN and SVM classifiers are used to classify the mammograms. Mammography Image Analysis society (MIAS) database is used to evaluate the proposed system. Experimental results show that the proposed system produces 93.39% classification accuracy to classify normal/abnormal images and 92.10% to classify benign/malignant images. Also the SVM classifier produces better accuracy than the KNN classifier.