LUCID: Learned Undersampling-Adaptive Consistency-Guided Inference with Deterministic Flow Matching for Sparse-View CT Reconstruction
2026-06-15 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors propose Lucid, a method to improve CT scans taken with fewer angles, which usually causes blurry images and missing details. They trained Lucid only on clear, high-quality CT images so it can adapt to different amounts of data when reconstructing scans without needing to be retrained. Lucid combines the limited-angle image and noise, then refines the image step-by-step while checking that it fits the original scan data, leading to clearer images with fewer fake structures. Their tests show Lucid works well across various scan settings, keeping image quality and structure more accurate than other methods.
Sparse-view CTAngular undersamplingFlow MatchingImage reconstructionGaussian distributionFiltered Back Projection (FBP)Data consistencyGenerative modelsRadiation dose reductionStructural fidelity
Authors
Jigang Duan, Jiayi Wang, Heran Wang, Ping Yang, Genwei Ma, Xing Zhao
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 prior for sparse-view CT. Lucid is trained only on high-quality CT images to learn a continuous transport between a Gaussian distribution and the high-quality CT image distribution, independent of view sampling. During inference, the sampling sparsity level is explicitly incorporated to adapt the generative trajectory of a single pretrained model. Specifically, Lucid constructs a degradation-matched initial state by sparsity-weighted fusion of the sparse-view FBP image and Gaussian noise, performs sparsity-modulated Flow Matching updates, and applies projection-domain data-consistency correction after each prior update. Experiments under multiple sparse-view settings show that Lucid achieves stable reconstruction performance across different sampling densities, improves image quality and structural fidelity, and reduces the risk of hallucination-like structures in generative sparse-view CT reconstruction.