A survey on different parametric based datasets, screening and deep learning strategies for diagnosis of breast carcinoma

Abstract

Breast carcinoma poses a serious concern to the public's health since it is the most usual type of tumor identified globally and the primary reason for women's cancer-related fatalities. As a result, early identification and medical intervention of malignant breast cancers can greatly improve patient outcomes and enable efficient therapy. Radiologists and physicians still struggle to provide accurate and consistent interpretations, which may result in misunderstandings and unnecessary biopsies. Many studies have investigated using Deep Learning (DL) techniques in conjunction with breast screening protocols to achieve accurate early detection of breast cancer. This review briefly presents the fundamental concepts of mammography and other screening methods and available datasets for deep learning in concern to breast cancer. This paper explores the usual recent advancements in Deep Learning algorithms systems used for breast tumor detection. Furthermore, it offers a succinct synopsis of datasets that are accessible to the public and investigates the most popular metrics for assessing computer-aided breast cancer diagnosis systems. Possible research directions in this emerging topic are finally described. In addition to providing a thorough analysis of the subject, this paper aims to encourage and point scientists, medical professionals, researchers, and other healthcare workers in the right direction when developing innovative applications for early breast cancer detection using Deep Learning.