A Hybrid Classifier for Mammography CAD

Abstract

Breast cancer is a very deadly disease for women. For the time being, mammographic screening remains the most effective method for early detection of breast cancer. However, reading mammography is a time-consume error-prone work. Therefore, many computer-aided detection and diagnosis systems (CAD) have been developed to assist radiologists in detecting and classifying mammographic mass. Most of those CAD system used single classifier for the classification of mass patterns into benign and malignant, or normal and mass or calcification. Increasing number of researches demonstrated that multi-classifier is an effective approach to improve the classification performance of CAD system. In this paper, we present a new hybrid classifier for mammographic CAD by hybridizing Logistic Regression (LR) and K-nearest neighbor (KNN) classifiers. To test and evaluate the proposed hybrid classifier, several experiments were carried out. The experimental results show that the proposed hybrid method achieves better performance then those two single classifiers (i.e., LR classifier and KNN classifier).