Self-supervision has demonstrated to be an effective learning strategy when training target tasks on small annotated data-sets. While current research focuses on creating novel pretext tasks to learn meaningful and reusable representations for the target task, these efforts obtain marginal performance gains compared to fully-supervised learning. Meanwhile, little attention has been given to study the robustness of networks trained in a self-supervised manner. In this work, we demonstrate that networks trained via self-supervised learning have superior robustness and generalizability compared to fully-supervised learning in the context of medical imaging. Our experiments on pneumonia detection in X-rays and multi-organ segmentation in CT yield consistent results exposing the hidden benefits of self-supervision for learning robust feature representations.
@article{arxiv.2105.06986,
title = {Evaluating the Robustness of Self-Supervised Learning in Medical Imaging},
author = {Fernando Navarro and Christopher Watanabe and Suprosanna Shit and Anjany Sekuboyina and Jan C. Peeken and Stephanie E. Combs and Bjoern H. Menze},
journal= {arXiv preprint arXiv:2105.06986},
year = {2021}
}