Classification of Breast Cancer using Neural Networks

Early detection is an important and promising medical activity to improve the chances of survival of the patients. Classification of Breast Cancer using various Neural Networks is performed and their results are compared in order to identify the network suitable for distinguishing between a benign and a malignant one. A Probabilistic Neural Network (PNN) for breast cancer classification producing accuracies up to 98% and a Back Propagation Neural Network (BPNN) having two output neurons are proposed and their accuracies are being compared with existing BPNN having one output neuron, the Radial Basis Function (RBF) Networks, the Learning Vector Quantization (LVQ) networks and also with Adaptive Neuro Fuzzy Inference System (ANFIS). Wisconsin Breast Cancer Diagnosis (WBCD) dataset is used for training and testing of the proposed neural networks.