Detection of Breast Cancer Using Deep Learning Techniques

Abstract

Evaluation of Histopathology images are a vital approach that is used for the breast cancer detection. To build up the efficiency of breast cancer detection and to reduce the burden of doctors and specialists, we layout various Deep Learning algorithms to recognize most cancers with the usage of histopathology scans. This paper follows several deep learning models like Convolutional Neural network (CNN) and Vgg16 for the recognition method. The dataset we used for class manner is Breast Histopathology Images which contain positive and negative images. We examined breast Histopathology images of 2,77,524 patients of which 198,748 images are IDC (-) and 78,786 images are IDC (+). This shows the deep learning algorithms can greatly facilitate the breast cancer detection, improving the accuracy and speed of detection. One of the most common cancers is Invasive Ductal Carcinoma (IDC). To determine the aggressiveness score to whole-mount specimen, doctors typically focus on areas containing IDC. Therefore, one of the common pre-processing steps for automatic aggressive categorization is to identify the exact region of IDC along the mounting side.