Re-identification from histopathology images

Jonathan Ganz, Jonas Ammeling, Samir Jabari, Katharina Breininger, Marc Aubreville

Abstract: Highlights•Patient re-identification from digital pathology slides is feasible, with limitations.•Higher accuracy is observed for slides originating from the same tumor.•Deep learning models for patient re-identification pose a threat to patient privacy.•Proposed a risk assessment scheme for evaluating patient privacy risks before publishing data.