Transforming Breast Cancer Identification: An In-Depth Examination of Advanced Machine Learning Models Applied to Histopathological Images

Abstract

Breast cancer stands as one of the most prevalent and perilous forms of cancer affecting both women and men. The detection and treatment of breast cancer benefit significantly from histopathological images, which carry crucial phenotypic information. To enhance accuracy in breast cancer detection, Deep Neural Networks (DNNs) are commonly utilized. Our research delves into the analysis of pre-trained deep transfer learning models, including ResNet50, ResNet101, VGG16, and VGG19, for identifying breast cancer using a dataset comprising 2453 histopathology images. The dataset categorizes images into two groups: those featuring invasive ductal carcinoma (IDC) and those without IDC. Through our analysis of transfer learning models, we observed that ResNet50 outperformed the other models, achieving impressive metrics such as accuracy rates of 92.2%, Area under Curve (AUC) rates of 91.0%, recall rates of 95.7%, and a minimal loss of 3.5%.