AI RESEARCH
MeDUET: Disentangled Unified Pretraining for 3D Medical Image Synthesis and Analysis
arXiv CS.CV
•
ArXi:2602.17901v2 Announce Type: replace-cross Self-supervised learning (SSL) and diffusion models have advanced representation learning and image synthesis, but in 3D medical imaging they are still largely used separately for analysis and synthesis, respectively. Unifying them is appealing but difficult, because multi-source data exhibit pronounced style shifts while downstream tasks rely primarily on anatomy, causing anatomical content and acquisition style to become entangled. In this paper, we propose MeDUET, a 3D Medical image Disentangled UnifiEd Pre.