Deep learning for clinical mammography screening

Abstract

Breast cancer is the most common cancer among women worldwide. Today, the vast majority of breast cancers are diagnosed from screening mammography. Multiple randomized clinical studies have demonstrated that screening mammography can help reduce the number of deaths from breast cancer among women ages 40 to 74, especially for those over age 50 [4], and can provide women diagnosed with breast cancer more options for less aggressive treatment [7]. Screening mammography is the first entry into the funnel of clinical mammography. A screening mammogram can result in a suspicious finding, leading the patient to receive additional imaging, and even surgical biopsy if the additional imaging. Screening mammography, as the first part of this funnel, is a place for machine learning to have impact on the largest amount of patients. In this work, we apply machine learning models to tasks in clinical mammography such as density estimation, and Bi-Rads prediction.