Prostate Cancer Classification Based on Best First Search and Taguchi Feature Selection Method

Prostate cancer is the second most common cancer occurring in men worldwide, about 1 in 41 men will die because of prostate cancer. Death rates of prostate cancer increases with age. Even though, it being a serious condition only about 1 man in 9 will be diagnosed with prostate cancer during his lifetime. Accurate and early diagnosis can help clinician to treat the cancer better and save lives. This paper proposes two phases feature selection method to enhance prostate cancer early diagnosis based on artificial neural network. In the first phase, Best First Search method is used to extract the relevant features from original dataset. In the second phase, Taguchi method is used to select the most important feature from the already extracted features from Best First Search method. A public available prostate cancer benchmark dataset is used for experiment, which contains two classes of data normal and abnormal. The proposed method outperforms other existing methods on prostate cancer benchmark dataset with classification accuracy of 98.6%. The proposed approach can help clinicians to reach at more accurate and early diagnosis of different stages of prostate cancer and so that they make most suitable treatment decision to save lives of patients and prevent death due to prostate cancer.