Lune

CVPR Findings2026

PTF-CT: Polar-Aware Temporal-Frequential Iterative Reconstruction for Sparse-View CT

Borui Kang, Guanyi Qin, Chuanpu Li, Yueming Jin

Abstract

Sparse-view computed tomography (CT) reconstruction remains fundamentally ill-posed: aggressively reducing projection views amplifies streak artifacts, destroys fine structures, and severely challenges generalization across sparsity and noise regimes. We propose Polar-aware Temporal-Frequential CT reconstruction (PTF-CT), a new structural formulation that unifies model-based iterative inversion with geometry-informed priors and progressive frequency curriculum learning. We instantiate the inversion loop using a lightweight Restormer-based U-shaped Transformer, enabling efficient long-range spatial modeling within each update while keeping memory and computation tractable. Beyond this backbone, PTF-CT introduces two complementary mechanisms: (1) a geometryderived, lightly parameterized Information Density Module that explicitly compensates under-informed radial regions dictated by projection sensitivity imbalance; and (2) a Multi-Scale Temporal-Frequential Module that progressively injects frequency components across resolutions and iterative stages, allowing reconstruction to evolve from coarse artifact suppression to fine structural refinement. Experiments on the AAPM primary benchmark, and generalization across DeepLesion, demonstrate consistent stateof-the-art PSNR/SSIM under 48/32/24-view regimes in both noiseless and noisy conditions. PTF-CT produces sharper, more structurally faithful reconstructions, and establishes a unified structural perspective for physically grounded deep iterative CT inversion.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Improving Diffusion Models for Inverse Problems using Manifold Constraints cites Solving Inverse Problems in Medical Imaging with Score-Based Generative Models
  • Diffusion Posterior Sampling for General Noisy Inverse Problems cites Solving Inverse Problems in Medical Imaging with Score-Based Generative Models
  • Diffusion Posterior Sampling for General Noisy Inverse Problems cites Improving Diffusion Models for Inverse Problems using Manifold Constraints
Builds onCited byLeft to right by year; the more cited, the higher.

Builds on8

  1. Restormer: Efficient Transformer for High-Resolution Image RestorationSyed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat et al.CVPR 2022 · 3,348 citations
  2. Improving Diffusion Models for Inverse Problems using Manifold ConstraintsHyungjin Chung, Byeongsu Sim, Dohoon Ryu, Jong Chul YeNeurIPS 2022 · 738 citations
  3. Solving Inverse Problems in Medical Imaging with Score-Based Generative ModelsYang Song, Liyue Shen, Lei Xing, Stefano ErmonICLR 2022 · 721 citations
  4. Diffusion Posterior Sampling for General Noisy Inverse ProblemsHyungjin Chung, Jeongsol Kim, Michael Thompson McCann, Marc Louis Klasky et al.ICLR 2023 · 152 citations
  5. Unsupervised Self-Prior Embedding Neural Representation for Iterative Sparse-View CT ReconstructionXuanyu Tian, Lixuan Chen, Qing Wu, Chenhe Du et al.AAAI 2025 · 4 citations

Related papers

  1. IntraTomo: Self-supervised Learning-based Tomography via Sinogram Synthesis and PredictionGuangming Zang, Ramzi Idoughi, Rui Li, Peter Wonka et al.ICCV 2021 · 128 citations
  2. Physics-Aware Accelerated Unrolling Model for Sparse-View CT ReconstructionShaojie Guo, Yingying Fang, Junkang Zhang, Yan WangAAAI 2026
  3. Learning Projection Views for Sparse-View CT ReconstructionLiutao Yang, Rongjun Ge, Shichang Feng, Daoqiang ZhangACM MM 2022 · 12 citations
  4. TPG-INR: Target Prior-Guided Implicit 3D CT Reconstruction for Enhanced Sparse-View ImagingQinglei Cao, Ziyao Tang, Xiaoqin TangICCV 2025 · 1 citation
  5. GGPT: Geometry-Grounded Point TransformerYutong Chen, Yiming Wang, Xucong Zhang, Sergey Prokudin et al.CVPR 2026 · 2 citations
Dusk over the sea between two cliffs drawn in fine vertical lines