Assessing the Ability of Generative Adversarial Networks to Learn Canonical Medical Image Statistics | IEEE Journals & Magazine | IEEE Xplore