LUCID: Learned Undersampling-Adaptive Consistency-Guided Inference with Deterministic Flow Matching for Sparse-View CT Reconstruction
arXiv:2606.16212v1 Announce Type: cross Abstract: Sparse-view CT reduces radiation dose and scanning time by acquiring fewer projection views, but angular undersampling makes reconstruction severely ill-posed, causing streak artifacts, structural blurring, and loss of fine details. Existing supervised methods are often tied to specific sampling settings, whereas generative methods may introduce anatomically inconsistent hallucination-like structures under severe undersampling. We propose Lucid, a sparsity-adaptive, consistency-guided reconstruction framework based on a Flow Matching generative
Advances in generative AI, particularly flow matching architectures, are enabling more robust solutions for inherently ill-posed inverse problems like sparse-view CT reconstruction.
This development represents a significant step towards practical, lower-radiation medical imaging, addressing critical limitations of current sparse-view methods.
Sparse-view CT scans can now potentially yield higher quality images with fewer artifacts and better detail, making them more diagnostically reliable across various sampling settings.
- · Medical imaging manufacturers
- · Hospitals and clinics
- · Patients needing CT scans
- · AI in healthcare companies
- · Traditional algorithmic CT reconstruction methods
- · Providers reliant on high-radiation CT protocols
Reduced radiation exposure for patients undergoing CT scans, leading to safer diagnostic procedures.
Increased adoption of sparse-view CT technologies due to improved image quality, potentially expanding access to diagnostics.
Further integration of advanced AI generative models into other medical imaging modalities, driving a broader paradigm shift in diagnostics.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI