Deep Learning Approach for Mammographic Breast Density Classification and Cancer Risk Prediction

Abstract

Breast cancer is one among the various vulnerable types of cancer, after skin cancer and lung cancer. Although deaths from breast cancer have decreased over the years, it is still the major leading causes of women deaths of all races. Many research efforts have been taken to prevent breast cancer by using different breast cancer biomarkers in the last few decades. “Mammographic Breast Density” is amongst various significant markers utilized for the prevention of breast cancer. As there is an increase in the “Mammographic Breast Density”, the mammograms sensitivity also decreases causing wrong prediction of breast cancer. The primal motive behind this article is to study all the research innovations for Mammographic Breast Density classification. This survey article covered all Deep Learning methods proposed for mammographic breast density classification. From 2010-2017 there is an inclination towards the Machine Learning approach, and from 2017 onwards, there is more research inclination towards the Deep Learning approach. Statistics of classification accuracy of Deep Learning is in between 86%-98.87%. Due to the variations, no methods were found so precise and accurate. Hence, current mammographic breast density assessment is subjective, thus raising the need to develop an accurate and accurate mammographic Breast Density classification tool suitable for clinical practice. Implementing a successful CAD system for breast density classification is a social need and can act as a supporting mechanism for the precise classification of mammograms. More research efforts are required in this area to reduce faulty predictions.