Identification of Breast Cancer Metastasis Using Boosting Algorithms on Cytopathologic Data

Authors : Safak KAYIKCI

Pages : 11-21

View : 16 | Download : 8

Publication Date : 2021-08-30

Article Type : Research Paper

Abstract :Breast cancer is the second most common cancer among women after lung cancer. Early diagnosis of cancer can positively affect the recovery process from disease. Several machine learning-based approaches have been studied for cancer detection on histopathological images. In this study, identification of cancer type has been made using Gradient Boosting Machine insert ignore into journalissuearticles values(GBM);, eXtreme Gradient Boost insert ignore into journalissuearticles values(XGBoost);, and Light Gradient Boosting Machine insert ignore into journalissuearticles values(LightGBM); algorithms. The performances of these techniques have been measured on the Breast Cancer Wisconsin insert ignore into journalissuearticles values(Diagnostic); dataset. According to the results obtained, Gradient Boosting Machine insert ignore into journalissuearticles values(GBM); got the highest accuracy rate with 97.02% success. Although there is no pathological prior knowledge about the disease, high success has been achieved in diagnosing with the deep learning architectures used.
Keywords : Breast cancer, eXtreme gradient boost, gradient boosting machine, light gradient boosting machine