Recent Advancement and Challenges of Deep Learning for Breast Mass Classification from Mammogram Images

Deep learning (DL) has become a critical component of medical image processing. Over time, DL methods have changed. Advances in the field of DL have resulted in a computer-aided diagnosis system (CADs) that is more sophisticated and self-reliant. In medical image analysis, convolutional neural networks (CNN) are becoming increasingly extensively employed as a DL approach. This study aims to survey state-of-the-art approaches of breast mass classification using CNNs. The breast cancer mammography repositories have also been examined. Various limitations that demand further examination are also discussed. We looked at articles published on well-known publishing platforms like Google Scholar, PubMed, Science Direct, and IEEE Xplore to conduct the literature study. These papers are all SCOPUS/SCI/SCIE indexed and focus on using CNN algorithms in mammogram images. We also present the advancements and challenges of CNNs for breast cancer diagnosis. When it comes to medical image processing, using CNNs has proved to be more beneficial to researchers than using a traditional approach. However, better architectures, larger datasets that address class imbalance issues and improved optimization methods are still required.