Enhancing Breast Cancer Recognition in Histopathological Imaging Using Fine-Tuned CNN

Abstract

Global Cancer Statistics reports that of the 2.3 million cases of breast cancer worldwide, 600,000 result in death. Factors contributing to breast cancer in women include both genetic and lifestyle influences. One method for recognizing breast cancer is through histopathology images. Recently, deep learning has gained significant attention in machine learning due to its powerful capabilities in modeling complex data, such as images. In this study, we classify breast cancer by training a Convolutional Neural Network (CNN) model on a dataset of histopathology images annotated and validated by experts, containing two classes. We propose an optimization strategy for CNN models to enhance breast cancer recognition performance, applying a fine-tuning strategy to MobileNetV2 and InceptionResNetV2 to evaluate CNN performance in classifying breast cancer within histopathological images. The experimental results demonstrate that the model achieves optimal performance with an accuracy of 96.22%.