Cross-Distribution Diffusion Priors-Driven Iterative Reconstruction for Sparse-View CT

2025-09-16

AI Summary

Sparse-view CT (SVCT) reconstruction reduces radiation dose and improves temporal resolution but suffers from view-missing artifacts and domain shifts caused by inter-scanner, inter-protocol, and inter-anatomical variations, leading to severe performance degradation in out-of-distribution (OOD) scenarios. To address this, we propose a robust reconstruction framework integrating diffusion priors with model-based iterative optimization. Our method introduces a Transformer-based scalable interpolation network that jointly learns domain-invariant and domain-specific features via classifier-free guidance and stochastic conditional dropout. It further combines physics-driven iterative reconstruction with denoising diffusion sampling. Evaluated on multi-center datasets, our approach significantly outperforms state-of-the-art methods, maintaining high-fidelity structural details and reconstruction stability even under OOD conditions. The framework demonstrates strong generalizability and clinical translatability.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Sparse-View CT (SVCT) reconstruction enhances temporal resolution and reduces radiation dose, yet its clinical use is hindered by artifacts due to view reduction and domain shifts from scanner, protocol, or anatomical variations, leading to performance degradation in out-of-distribution (OOD) scenarios. In this work, we propose a Cross-Distribution Diffusion Priors-Driven Iterative Reconstruction (CDPIR) framework to tackle the OOD problem in SVCT. CDPIR integrates cross-distribution diffusion priors, derived from a Scalable Interpolant Transformer (SiT), with model-based iterative reconstruction methods. Specifically, we train a SiT backbone, an extension of the Diffusion Transformer (DiT) architecture, to establish a unified stochastic interpolant framework, leveraging Classifier-Free Guidance (CFG) across multiple datasets. By randomly dropping the conditioning with a null embedding during training, the model learns both domain-specific and domain-invariant priors, enhancing generalizability. During sampling, the globally sensitive transformer-based diffusion model exploits the cross-distribution prior within the unified stochastic interpolant framework, enabling flexible and stable control over multi-distribution-to-noise interpolation paths and decoupled sampling strategies, thereby improving adaptation to OOD reconstruction. By alternating between data fidelity and sampling updates, our model achieves state-of-the-art performance with superior detail preservation in SVCT reconstructions. Extensive experiments demonstrate that CDPIR significantly outperforms existing approaches, particularly under OOD conditions, highlighting its robustness and potential clinical value in challenging imaging scenarios.

Problem

• Addressing artifacts in sparse-view CT from view reduction
• Mitigating domain shifts from scanner and protocol variations
• Improving out-of-distribution reconstruction performance in CT

Innovation

• Cross-distribution diffusion priors for iterative reconstruction
• Unified stochastic interpolant framework with scalable transformer
• Classifier-free guidance across multiple datasets for generalization