Comparison of Two Different Deep Learning Architectures on Breast Cancer

Abstract

Breast cancer is one of the diseases becoming widespread gradually nowadays. Diagnosis and treatment of breast cancer are performed by some specialist doctors. Timely and accurate detection of this disease is lifesaving. DenseNet-201 and Xception deep learning architectures are used in this study. The performance of these two different deep learning methods are evaluated on the breast cancer dataset. The dataset consists of some benign and malignant cancer images. There are 20748 images for training and 5913 images for testing. According to the results obtained, DenseNet-201 method reaches an F-1 accuracy score of 92.24%, and the Xception method achieves an F-1 accuracy score of 92.41% when trained on the used dataset.