A Systematic Review on Deep Convolutional Neural Network-based Breast Cancer Classification on Histopathological Images

Abstract

Breast cancer (BC) is an increasingly prevalent malignant disease in females globally. Lately, early diagnoses and the best adjuvant therapy have considerably enhanced patient outcomes. However, new challenges have occurred as our growing understanding of tumors has exposed their complex nature, and diagnoses by histopathology have proved to be helpful in guiding BC treatment. We faced an absolute necessity for accurate histopathologic BC diagnoses to make better therapy decisions as patient demand for personalized BC therapy increases. Furthermore, current development in memory capacity and computational power resulted in the applications of medical image processing and Deep Learning (DL) techniques to analyze and process histopathological images (HIs) of BC. Therefore, this study performs a Systematic Review of Deep Convolutional Neural Network based BC Classification on HIs. This survey aims to review the conventional and recently developed DL techniques for BC diagnosis using HIs. Firstly, the role of machine learning (ML) and DL algorithms for HI classification for BC detection is elaborated briefly. Next, the recently developed DL-based HI classification models for BC are reviewed in detail. Moreover, a comparison stud of the reviewed models with result analysis is performed. Furthermore, an elaborate description of the challenging issues with possible future directions is identified at the end of the survey.