用于极端CT重建的前馈分层高斯扩散

中文摘要 AI 辅助

从严重受限的投影数据重建三维计算机断层扫描（CT）是一个高度不适定的问题。稀疏角度采样、受限的角度覆盖范围以及低光子计数可能单独或共同出现，导致全局解剖结构和局部组织细节模糊不清。许多基于学习的CT重建方法针对单一主要退化情况设计。现有的扩散模型和高斯方法通常在共享表示中恢复全局结构和局部细节。我们提出HiGDiff，一种前馈分层高斯扩散框架，该框架在空间上以及从结构到细节两个维度分解重建过程。基于物理条件的解剖锚点和前景容量场将可学习的高斯基元分配到信息丰富的区域。结构扩散阶段首先恢复全局衰减几何结构，其学习到的表示为细节扩散阶段提供条件，以处理剩余边界和组织过渡。生成的高斯库被渲染为衰减场，并通过梯度隔离残差模块进一步优化。在三个不同的CT基准数据集上进行的实验表明，该方法在单独、配对和联合退化设置下均实现了最先进的重建性能，包括在低剂量CT图像和投影数据（LDCT-PD）集合上的宏观平均峰值信噪比（PSNR）提升5.81 dB，结构相似性指数测量（SSIM）提升0.113。代码和实验配置可在此https URL公开获取。

英文摘要

Reconstructing three-dimensional computed tomography (CT) from severely constrained projections is highly ill-posed. Sparse angular sampling, restricted angular coverage, and low photon counts can occur individually or jointly, obscuring global anatomy and local tissue detail. Many learned CT reconstruction methods are tailored to a single dominant degradation. Existing diffusion and Gaussian approaches commonly recover global structure and local detail within a shared representation. We propose HiGDiff, a feed-forward hierarchical Gaussian diffusion framework that decomposes reconstruction both spatially and from structure to detail. Physics-conditioned anatomical anchors and a foreground capacity field allocate learnable Gaussian primitives to informative regions. A structure diffusion stage first recovers global attenuation geometry, and its learned representation conditions a detail diffusion stage for residual boundaries and tissue transitions. The resulting Gaussian banks are rendered as attenuation fields and further refined by a gradient-isolated residual module. Experiments on three distinct CT benchmark datasets demonstrate state-of-the-art reconstruction performance across isolated, paired, and joint degradation settings, including improvements of 5.81 dB in macro-average peak signal-to-noise ratio (PSNR) and 0.113 in structural similarity index measure (SSIM) on the Low Dose CT Image and Projection Data (LDCT-PD) collection. Code and experimental configurations are openly available at https://github.com/Bean-Young/HiGDiff.

发表机构

• University of Alberta（阿尔伯塔大学）
• University of Sydney（悉尼大学）

机构由 AI 辅助整理，请以论文原文为准。