Reproduce “Merlin Plus: A Large-Scale, Multi-Cancer, Image-Mask-Report Dataset” 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
Pedro R. A. S. Bassi, Wenxuan Li, Szymon Plotka, Ruby Honjol, Jakub Przado, Xinze Zhou, Kang Wang, Yang Yang and 5 more
Merlin Plus extends a 25,494-scan CT-and-report collection with 1,153 radiologist-verified tumor masks across nine organs. Its report-guided workflow generates initial masks, prioritizes low-confidence cases for radiologist review, and retrains with corrected masks and reports. On external validation, R-Super achieved average tumor-detection F1 of 79, versus 57 for a Merlin-trained classifier without tumor-mask supervision; this supports mask-augmented training, not clinical benefit.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.