Injecting and removing suspicious features in breast imaging with CycleGAN: A pilot study of automated adversarial attacks using neural networks on small images.

Journal: European journal of radiology

Published Date: Sep 10, 2019

Abstract

PURPOSE: To train a CycleGAN on downscaled versions of mammographic data to artificially inject or remove suspicious features, and to determine whether these AI-mediated attacks can be detected by radiologists.

Authors

Anton S Becker

From the Institute of Diagnostic and Interventional Radiology, University Hospital Zurich, Zurich, Switzerland.
Lukas Jendele

Department of Software Engineering, Faculty of Mathematics and Physics, Charles University, Czech Republic.
Ondrej Skopek

Department of Computer Science, ETH Zurich, Zurich, Switzerland.
Nicole Berger

Institute of Diagnostic and Interventional Radiology, University Hospital Zurich, Zürich, Switzerland.
Soleen Ghafoor
Magda Marcon
Ender Konukoglu

Keywords

Area Under Curve Breast Breast Neoplasms Case-Control Studies Female Humans Image Interpretation, Computer-Assisted Mammography Neural Networks, Computer Pilot Projects Radiologists ROC Curve

External Resources

View on PubMed Access via DOI PubMed (31539791)

Injecting and removing suspicious features in breast imaging with CycleGAN: A pilot study of automated adversarial attacks using neural networks on small images.

Abstract

Authors

Keywords

External Resources

Popular Topics

Recent Journals