Breast diagnostic based on mammography data using deep learning

Resumo

This work investigates the usage of a deep learning model to support breast diagnostics from mammographic image data [1]. We are collaborating as a beta-tester on the NYU Breast Cancer Classification Meta-repository. Medical images were obtained for testing from a Brazilian Federal Hospital - HUCFF and a database at IEN have been created for research continuity. The main objective is to evaluate a model that could improve radiologist clinical routine, a computational system (Figure 1) for breast cancer diagnostic for standardization and improvement of deep learning in breast cancer for data owners and developers that has the entire pipeline encapsulated in a Docker container [2]. It includes ready-to-use Docker images of screening mammography models.