PHASOR: Anatomy- and Phase-Consistent Volumetric Diffusion for CT Virtual Contrast Enhancement

PHASOR is a pre-publication research paper (5 days old, 0 stars/forks) with no released code, community, or adoption signals. The core contribution—anatomy-consistent volumetric diffusion for CT virtual contrast enhancement—combines existing techniques (diffusion models + 3D medical imaging) in a domain-specific way. The novelty is incremental-to-moderate: applying phase-consistency constraints and volumetric diffusion to an existing problem (VCE from NCCT) rather than introducing a fundamental new technique. As a reference implementation only (no released artifact), it has zero defensibility—reproduction depends entirely on the paper's reproducibility and whether authors release code. Frontier risk is HIGH because: (1) medical imaging is a core capability area for frontier labs (OpenAI/Anthropic via partnerships, Google/Meta via health initiatives); (2) virtual contrast synthesis directly competes with their synthetic imaging pipelines; (3) diffusion models are commoditized infrastructure; (4) the domain-specific contribution (phase consistency for anatomy) is algorithmic and incremental, not architectural—easily integrated into existing medical imaging products. A frontier lab could implement this as a feature module in 2-4 weeks. The paper may be technically sound but lacks the moat (no users, no ecosystem, no code release) to survive competitive pressure from well-resourced medical AI teams.