Application of GANs to Augment the Mammography Repository

Abstract

Generative adversarial networks (GANs) offer an innovative approach to synthetic image generation. They have significantly impacted the creation of images that would otherwise be difficult to obtain. In this study, we examine several GAN architectures to determine whether they can generate synthetic mammography images to enrich an existing repository, thereby improving AI training for breast-cancer detection and supporting research into this disease with a more diverse dataset.