Scholarlink

Computed Tomography (CT) reconstruction under sparse-view settings is crucial for reducing radiation exposure but often suffers from severe artifacts. In this paper, we reformulate sparse-view CT reconstruction as a measurement-domain inpainting task and introduce a score-based diffusion model that exploits structural properties of the sinogram. Our method achieves \(14\%\) higher PSNR and noticeable improvements in SSIM over existing approaches, demonstrating its effectiveness in producing reliable low-dose CT reconstructions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Solving Low-Dose Computer Tomography Inverse Problem by Learning the First-Order Score of the Sparse Sinogram Samples’ Distribution

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yuchen Quan,
  • Yaru Xue,
  • Haisu Zhu,
  • Yuzhu Gu,
  • Jing Li,
  • Yu Yan

摘要

Computed Tomography (CT) reconstruction under sparse-view settings is crucial for reducing radiation exposure but often suffers from severe artifacts. In this paper, we reformulate sparse-view CT reconstruction as a measurement-domain inpainting task and introduce a score-based diffusion model that exploits structural properties of the sinogram. Our method achieves \(14\%\) higher PSNR and noticeable improvements in SSIM over existing approaches, demonstrating its effectiveness in producing reliable low-dose CT reconstructions.