Roman C Maron, Sarah Haggenmüller, Christof von Kalle, Jochen S Utikal, Friedegund Meier, Frank F Gellrich, Axel Hauschild, Lars E French, Max Schlaak, Kamran Ghoreschi, Heinz Kutzner, Markus V Heppt, Sebastian Haferkamp, Wiebke Sondermann, Dirk Schadendorf, Bastian Schilling, Achim Hekler, Eva Krieghoff-Henning, Jakob N Kather, Stefan Fröhling, Daniel B Lipka, Titus J Brinker
BACKGROUND: A basic requirement for artificial intelligence (AI)-based image analysis systems, which are to be integrated into clinical practice, is a high robustness. Minor changes in how those images are acquired, for example, during routine skin cancer screening, should not change the diagnosis of such assistance systems. OBJECTIVE: To quantify to what extent minor image perturbations affect the convolutional neural network (CNN)-mediated skin lesion classification and to evaluate three possible solutions for this problem (additional data augmentation, test-time augmentation, anti-aliasing)...
March 2021: European Journal of Cancer