Reproduce “SciFusion: Physics-Guided Compositional Diffusion for Multimodal 3D Scientific Image Generation” 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
Luoxi Wang, Alessio Quercia, Hanno Scharr, Arya Bangun
SciFusion presents a shared volumetric generator for MRI, CT, and electron microscopy. It conditions generation on domain, object, and contrast factors, and adds modality-specific physics-inspired training guidance. Across 15 modalities, its I3D-FVD is 9.90, versus 11.80 for flattened conditioning. In zero-shot tests, it recombines familiar factors for two held-out MRI combinations, not unseen imaging factors.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.