Piecewise Dynamic Diffusion Regularization for Reconstruction of Cardiac Cine MRI

2026-07-03

AI Summary

This work addresses the challenging problem of reconstructing real-time cardiac cine MRI under free breathing, where severe undersampling and motion artifacts degrade image quality. The authors propose a variational reconstruction framework that integrates spatiotemporal diffusion priors, featuring a dedicated spatial layer to encode anatomical structure and a temporal layer to model cardiac motion. A novel piecewise dynamic diffusion regularization strategy is introduced to simultaneously preserve anatomical and motion consistency while substantially improving computational efficiency, enabling scalable reconstruction of long temporal sequences. Experimental results demonstrate that the proposed method outperforms existing classical, unsupervised, and diffusion-based approaches in both retrospectively accelerated and prospectively acquired real-time imaging scenarios, achieving significant gains in both reconstruction fidelity and computational speed.

Abstract

Real-time cardiac cine MRI enables visualization of the beating heart during free breathing, but severe undersampling and motion make reconstruction highly challenging. A central challenge for reconstruction is incorporating powerful priors of cardiac anatomy while remaining computationally efficient. We propose Piecewise Dynamic Diffusion Regularization (PDDR), a reconstruction method that integrates a spatiotemporal diffusion model as a generative prior within a variational reconstruction framework for cine MRI. The model employs dedicated spatial layers to encode anatomical structure and temporal layers to capture cardiac motion learned from gated cine data. PDDR leverages the dynamic prior in a piecewise manner, enabling the efficient use of spatiotemporal diffusion models for processing of long real-time sequences. Experiments on retrospectively accelerated and prospective real-time cine MRI demonstrate that PDDR outperforms classical, unsupervised, and diffusion-based methods, delivering high-quality reconstructions with substantially reduced computation time compared to state-of-the-art baselines. These results highlight PDDR as a practical and scalable solution for free-breathing, real-time cardiac MRI. Code is available at https://github.com/MLI-lab/pddr.

Problem

• cardiac cine MRI
• undersampling
• motion artifacts
• image reconstruction
• real-time MRI

Innovation

• diffusion model
• spatiotemporal prior
• piecewise regularization
• real-time cardiac MRI
• variational reconstruction