Machine learning approaches for breast cancer diagnosis and their comparison

Abstract

Machine learning approaches are used for building systems that can solve various diagnostic problems. Since breast cancer is highly incident on women, there is a need for such systems. Mammograms are used for early detection of breast cancer. The breast cancer diagnostic system extracts features from these mammograms and classifies them as malignant or benign. These systems are very helpful to doctors in detecting and diagnosing the disease faster than any other traditional methods. In this thesis an attempt has been made to classify the extracted features from mammograms as benign or malignant by using Naive Bayes, K-NN, Multilayer Perceptron, Radial Basis Function Networks, Support Vector Machine approaches. Performance variation of the approaches by varying various parameters is studied. Finally the results are compared to find the best performing approaches.