Abstract
The breast cancer is one of the wide spread diseases around the world. Cancer develops in a milk duct and then spreads to the surrounding breast tissues. This initial stage of progression is called invasive ductal carcinomas (IDC). Almost 80% of all breast cancers are invasive ductal carcinomas. If IDC is detected at early stages, the patient can be treated and will have a high survival rate, whereas undetected cancer may spread into other parts of the body, as well as surrounding breast tissues. In this work, the dataset that contains breast cancer histopathology images was used. The objective of this work is to implement a convolutional neural network (CNN) model for accurate IDC classification, by balancing the dataset and tuning hyperparameters. The proposed model achieves an accuracy of 92% for the classification of histopathological images, and outperforms the baseline CancerNet model with accuracy of 86%. Furthermore, our experimental results demonstrate the superiority of our approach over the pre-trained networks, such as VGG16, DenseNet and ResNet18.