Responsible AI Seminar: Sharing synthetic medical images - a way to circumvent GDPR?
On 17 September, Anders Eklund, Linköping University, will give a talk for the inter-university seminar series on responsible AI.
Sharing synthetic medical images - a way to circumvent GDPR?
Deep learning research in medical imaging is currently impeded by the lack of available training data. There is not a lack of data per se, but the major part of medical images stored in hospital databases cannot be used for research, due to perceived and actual regulatory barriers. Recent research has shown that generative adversarial networks (GANs) can be trained to synthesize very realistic images (e.g. thispersondoesnotexist.com). In this presentation I will focus on how GANs can be used for synthesizing medical images, 2D and 3D, and how this relates to data sharing. Can synthetic medical images be seen as anonymized, as they do not belong to a specific person, and therefore be shared freely? Can AI models be successfully trained using synthetic images?
About the Responsible AI Seminar Series
Responsible AI draws on widely different scientific disciplines, from the technical aspects of AI, via ethics, philosophy and law, to the individual realities of different application domains. We wish to take advantage of the limitations imposed by Covid-19 to start an informal conversation across Denmark about different aspects of Responsible AI via a hybrid format seminar series. We hope to catch your interest with three seminar talks before the summer vacation, following which we hope to merge efforts across universities to start a truly inter-university seminar series. Initiators: Aasa Feragen, Melanie Ganz and Sune Hannibal Holm from the DFF-funded project Bias and Fairness in Medicine.