Reproduce “ReSeg: Retraining-free segmentation of abdominal organs via generalizable semantic representation learning” 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
Derong Yu, Guoyan Zheng
ReSeg trains a model to segment the liver, spleen, and both kidneys across imaging domains without target-dataset retraining. It first learns appearance- and context-tolerant features from paired synthetic images, then trains segmentation on synthetic images generated from patient-derived organ masks. On AutoPET-Organ, it achieves 72.40% average Dice, versus 16.68% for ZeroShotSAM; this result covers four organs on the tested dataset.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.