An Ensemble Deep Learning Approach for Automated Breast Cancer Classification Using Histopathology Images

Breast cancer is a common worldwide risk to women's lives. This disease affects the health of millions of individuals globally. Early detection, therapy, and appropriate confinement of female patients are possible approaches to minimize breast cancer worldwide. Automated detection using histopathology microscopic images reduces the clinician's burden and saves thousands of people. This study proposed an ensemble approach for automated breast cancer classification using microscopic images. In this work, a publicly accessible microscopic image dataset containing 2480 Benign and 5429 Malignant microscopic images and three pre-trained DL models, namely, ResNet50, DenseNet169, and VGG19, are utilized. The accuracy of 96.88% is achieved by the ensemble model using transfer-learned DL models. This study highlights the working of ensemble approach to assist pathologists and enhance early detection of BC through automated analysis of histopathological images.