Development of Breast Cancer Detection Model Using Transfer Learning of Residual Neural Network (ResNet-50)

Abstract

Breast cancer disease is one of the most common and dangerous as well as being considered as the second most common world cause of cancer death in women. However, the early diagnostics and detection can provide a significant chance for correct treatment and survival. One of the most powerful tools that have shown extraordinary and superior results is the deep convolutional neural network. In this work, we propose an accurate and inclusive computational breast cancer diagnosis framework using ResNet-50 convolutional neural network to classify histopathological microscopy images. The proposed model employs transfer learning technique of the powerful ResNet-50 CNN pretrained on ImageNet to train and classify BreakHis dataset into benign or malignant. The simulation results showed that our proposed model achieves exceptional classification accuracy of 99% outperforming other compared models trained on the same dataset. Based on our novel approach, earlier detection to breast cancer as whether it is being benign or malignant can be stimulated and classified and thus save life and efforts.