Evaluation of GANs in the medical field

One of the major issues in the AI for medicine domain is the restricted availability of data. In this context generative approaches, such as Generative Adversarial Networks (GANs), can have major impact as data augmentation methodologies. However a main concern in using synthetic generated data is the problem of evaluating their quality. This talk presents and introduction to GANs and their evaluation, with a particular focus on the medical domain, where such evaluation is critical and particularly challenging.