Semantic Segmentation in Immunohistochemistry Breast Cancer Image using Deep Learning

Abstract

Cancer is affecting many people's lives. Uncontrollable growth of cells causes cancer. Thus, there is a need to find accurate results for the treatment of a patient by using the proper computation method. Immunohistochemistry (IHC) image is the study of stained cancerous tissue at a microscopic level. For IHC images, cancerous tissue taken during a biopsy is inspected to identify the protein expression region. Pathologists inspect the stained specimen and manually find the Region of Interest (ROI) and quantify the protein (brown color), which is subjective and time-consuming. Thus, there is a need to develop an assistive system to segment the protein expression region. In this paper, we segment the protein expression in the membranous region - HER2 expression of Breast Cancer using various segmentation models like FCN, SegNet and U-Net. U-Net performed well with an accuracy of 94%.