Extracting Rules from Feed Forward Neural Networks for Diagnosing Breast Cancer

Medical diagnosis is one of the major problems in medical application. This includes the limitation of human expertise to diagnose disease manually. Extensive amount of knowledge and data stored in medical databases require specialized tools for analysis and effective usage of data. Breast cancer is the largest cause of cancer deaths among women. The proposed method uses Feed Forward Neural Network (FFNN) model to diagnose breast cancer. The  performance of the neural network is increased by 98%. The diagnosis of breast cancer is examined on the well known and widely accepted Wisconsin breast cancer data (WBCD). The proposed method classifies whether a tumor is benign or malignant by five rules using Back propagation algorithm.