Diffusion-Based Limited-Angle CT Reconstruction under Noisy Conditions

TL;DR

This paper introduces a diffusion-based approach with a noise-aware mechanism for limited-angle CT reconstruction, effectively handling noisy measurements and improving image quality in challenging inverse problems.

Contribution

It proposes a novel diffusion framework with a noise-aware rectification mechanism for robust limited-angle CT reconstruction under noisy conditions.

Findings

Outperforms baseline models in data consistency.

Achieves higher perceptual quality in reconstructed images.

Generalizes well across different noise levels.

Abstract

Limited-Angle Computed Tomography (LACT) is a challenging inverse problem where missing angular projections lead to incomplete sinograms and severe artifacts in the reconstructed images. While recent learning-based methods have demonstrated effectiveness, most of them assume ideal, noise-free measurements and fail to address the impact of measurement noise. To overcome this limitation, we treat LACT as a sinogram inpainting task and propose a diffusion-based framework that completes missing angular views using a Mean-Reverting Stochastic Differential Equation (MR-SDE) formulation. To improve robustness under realistic noise, we propose RNSD$^+$, a novel noise-aware rectification mechanism that explicitly models inference-time uncertainty, enabling reliable and robust reconstruction. Extensive experiments demonstrate that our method consistently surpasses baseline models in data consistency and perceptual quality, and generalizes well across varying noise intensity and acquisition scenarios.