Analysis of different types of entropy measures for breast cancer diagnosis using ensemble classification

Breast cancer is a serious problem and common form of cancer diagnosed in the woman. Computer Aided Diagnosis (CAD) is a tool which can assist the radiologists in the detection of abnormalities in medical images. In this study, a CAD system for breast cancer using X-ray mammography is presented with a high level of sensitivity by wavelet entropy features. Discrete Wavelet Transform (DWT) of a digital mammogram provides a multi-resolution representation of it. The characteristics of a mammogram at different resolution levels are represented by computing wavelet entropy and used as features for the corresponding mammogram. Then, ensemble classification using K-Nearest Neighbors (KNN), Bayes, and Support Vector Machine (SVM) is employed to classify the abnormalities as benign/ malignant. The experiments show promising results with the high level of sensitivity and hence it is feasible for mammogram classification.