Digital Mammography DREAM Challenge: Challenging the Infrastructure with Deep Learning Models

Thomas Schaffter

09:50 - 10:05 | Monday 5 March 2018 | Treasure Island E


Breast cancer is the most frequently diagnosed solid cancer and second leading cause of cancer death among U.S. women. Screening mammography has been shown to decrease breast cancer-related mortality. The goal of the Digital Mammography DREAM Challenge was to apply an open science, crowdsourced approach to foster the development of machine learning approaches that can help decrease the recall rate of screening mammography, with the aim to shift the balance of routine breast cancer screening towards more benefit and less harm. Here I will describe the "challenges" that we have faced and the cloud infrastructure that we have developed to enable researchers to train models without accessing directly biomedical images.