The paper proposes SCMA, a structure-conditioned and metal-aware Flow Matching framework for CT metal artifact reduction. It uses a linearly interpolated, corrected image and the intermediate state as sample-specific structural guidance, while introducing time-varying spatial weights derived from the metal mask and its distance transform. During inference, conditional Flow Matching updates alternate with projection-consistency correction, allowing reliable measurements outside metal traces to constrain reconstruction. According to the abstract, experiments on simulated and real CT data show improved artifact suppression, local anatomy preservation, and fewer hallucination-like structures inconsistent with measured projections.
No heat snapshots are available in the last 24 hours.