Evaluating the Robustness of Self-Supervised Learning in Medical Imaging

Abstract

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.

References in corpus (5)

• Adam: A Method for Stochastic Optimization
• Validation, comparison, and combination of algorithms for automatic detection of pulmonary nodules in computed tomography images: the LUNA16 challenge
• The Liver Tumor Segmentation Benchmark (LiTS)
• Using Pre-Training Can Improve Model Robustness and Uncertainty
• The KiTS19 Challenge Data: 300 Kidney Tumor Cases with Clinical Context, CT Semantic Segmentations, and Surgical Outcomes

Cited by in corpus (4)

• A Review of Predictive and Contrastive Self-supervised Learning for Medical Images
• Focused Decoding Enables 3D Anatomical Detection by Transformers
• Robust and Explainable Framework to Address Data Scarcity in Diagnostic Imaging
• On the Robustness of Pretraining and Self-Supervision for a Deep Learning-based Analysis of Diabetic Retinopathy