Reproduce “Robust and Efficient Medical Imaging with Self-Supervision” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shekoofeh Azizi, Laura Culp, Jan Freyberg, Basil Mustafa, Sebastien Baur, Simon Kornblith, Ting Chen, Patricia MacWilliams and 26 more
REMEDIS combines supervised natural-image pretraining with self-supervised learning on unlabeled medical images, then fine-tunes on labeled task data. For lymph-node metastasis detection, its OOD test AUC was 0.876, versus 0.791 for the JFT-300M baseline without OOD training labels. It matched the baseline’s full-data OOD performance using 6% of those labels in retrospective simulated-shift experiments, not prospective clinical deployment.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.