Advancing Limited-Angle CT Reconstruction Through Diffusion-Based Sinogram Completion

TL;DR

This paper introduces a diffusion-based sinogram inpainting method for limited-angle CT that improves reconstruction quality by filling missing data at the projection level, leading to state-of-the-art results.

Contribution

It proposes a novel diffusion model approach using MR-SDEs for sinogram completion, accelerating the process with distillation and pseudo-inverse constraints.

Findings

Achieves state-of-the-art reconstruction quality.

Effectively suppresses artifacts and preserves details.

Demonstrates superior performance in scientific and clinical tests.

Abstract

Limited Angle Computed Tomography (LACT) often faces significant challenges due to missing angular information. Unlike previous methods that operate in the image domain, we propose a new method that focuses on sinogram inpainting. We leverage MR-SDEs, a variant of diffusion models that characterize the diffusion process with mean-reverting stochastic differential equations, to fill in missing angular data at the projection level. Furthermore, by combining distillation with constraining the output of the model using the pseudo-inverse of the inpainting matrix, the diffusion process is accelerated and done in a step, enabling efficient and accurate sinogram completion. A subsequent post-processing module back-projects the inpainted sinogram into the image domain and further refines the reconstruction, effectively suppressing artifacts while preserving critical structural details. Quantitative experimental results demonstrate that the proposed method achieves state-of-the-art performance in both perceptual and fidelity quality, offering a promising solution for LACT reconstruction in scientific and clinical applications.