Reproduce “3D Diffusion Models for Medical Imaging” 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
Frank-Peter Schilling, Daniel Barco, Philipp Denzel, Marc Stadelmann
The authors review diffusion and flow-based methods for generating three-dimensional medical images, focusing on how compressed latent representations make high-resolution volumes more tractable. They explain why slice-wise generation can disrupt anatomical continuity and why full-volume modeling remains computationally demanding. In a reported CT comparison, MAISI-v2 has FID 2.32 versus HA-GAN’s 13.60; these results are protocol-specific, and clinical reliability remains unresolved.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.