Mammogram Classification Using Maximum Difference Feature Selection Method

This paper developed a CAD (Computer Aided Diagnosis) system based on neural network and a proposed feature selection method. The proposed feature selection method is Maximum Difference Feature Selection (MDFS). Digital mammography is reliable method for early detection of breast cancer. The most important step in breast cancer diagnosis is feature selection. Computer automated feature selection is reliable and also it helps to improve the classification accuracy. GLCM (Gray Level Co-occurrence Matrix) features are extracted from the mammogram. The extracted features are selected based on a proposed MDFS method. Experiments have been conducted on datasets from DDSM (Digital database for Screening Mammography) database. Several feature selection methods are available. The accuracy of the model depends on the relevant feature selection. The proposed MDFS method selects only essential features and eliminates the irrelevant features. The experiment results show that neural network based model with proposed feature selection method improved the classification accuracy.