Fire ant optimisation feature selection method for breast cancer prediction

Abstract

Breast cancer is common disease in the today's world and many techniques are used to extract the cancer cell from the breast image. Most of the systems use features extracted from the images and these images are selected using feature selection techniques. The feature selection techniques help to a greater level in removal of irrelevant data from huge amount of data and fine tune the identification process and accuracy of relevant data. But still the prediction accuracy and number of the features selection factors are not yielding 100% prediction result. In this work, Nearest Density Fire Ant (NDFA) is proposed for breast cancer prediction and used for feature selection techniques. This technique is used for diagnosing and producing results compared to the previous methods such as Random Forest, Ant Colony and Genetic Algorithm. The proposed techniques are inspected on Wisconsin Breast Cancer Database (WBCD) and Breast Cancer Wisconsin WDBC data sets. The experimental result shows that proposed NDFA using Fire Ant Optimisation (FAO) technique produces better results in prediction and diagnosing of breast cancer.