Reproduce “HistoTx: Early fusion of H&E images and spatial transcriptomics at varying spatial transcriptomics resolution with self-supervised 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
Alice Driessen, Emil Malmsten, Steffen Wolf, Bernd Illing, Isinsu Katircioglu, Adriano Martinelli, Rolf Stahel, Solange Peters and 6 more
HistoTx combines spatial transcript measurements and H&E patches as tokens in a shared transformer, using molecular inputs when available and images alone otherwise. Training uses cross-modal consistency to transfer transcript-informed information to image representations. On HEST-Bench, it scores 0.38 for image-only gene-expression prediction versus 0.31 for the baseline; gains may partly reflect extra HEST-image exposure, and classification does not always improve.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.