Abstract:Diffusion models have recently demonstrated considerable advancement in the generation and reconstruction of magnetic resonance imaging (MRI) data. These models exhibit great potential in handling unsampled data and reducing noise, highlighting their promise as generative models. However, their application in dynamic MRI remains relatively underexplored. This is primarily due to the substantial amount of fully-sampled data typically required for training, which is difficult to obtain in dynamic MRI due to its spatio-temporal complexity and high acquisition costs. To address this challenge, we propose a dynamic MRI reconstruction method based on a time-interleaved acquisition scheme, termed the Glob-al-to-local Diffusion Model. Specifically, fully encoded full-resolution reference data are constructed by merging under-sampled k-space data from adjacent time frames, generating two distinct bulk training datasets for global and local models. The global-to-local diffusion framework alternately optimizes global information and local image details, enabling zero-shot reconstruction. Extensive experiments demonstrate that the proposed method performs well in terms of noise reduction and detail preservation, achieving reconstruction quality comparable to that of supervised approaches.

What problem does this paper attempt to address?

This paper attempts to address the image reconstruction problem in dynamic MRI. Specifically, the paper focuses on the following major challenges: 1. **Sparse Data Sampling**: In dynamic MRI, due to the limitations of time resolution and motion artifacts, only limited k-space data can usually be acquired. This leads to a decline in image reconstruction quality. 2. **Insufficient Training Data**: Existing diffusion model-based methods require a large amount of fully sampled data for training, but in practical applications, obtaining such data is very difficult and costly. 3. **Insufficient Detail Recovery**: Existing methods are inadequate in recovering fine structures and details of images, especially in dynamic MRI. To address these issues, the paper proposes a Global-to-local Diffusion Model (GLDM) based on a time-interleaved acquisition scheme. The main innovations of this method include: 1. **Zero-sample Learning**: By generating fully encoded training data through a time-interleaved acquisition scheme, it eliminates the reliance on a large amount of fully sampled data, thus achieving zero-sample learning. 2. **Two-stage Iterative Optimization**: Adopting a global and local diffusion model framework, it alternately optimizes the global structure and local details of the image, combining low-rank constraints and data consistency modules to enhance the robustness and reconstruction performance of the model. 3. **Time-interleaved Acquisition Scheme**: By merging data from adjacent time frames to generate fully encoded reference data, it improves the quality and efficiency of image reconstruction while maintaining time resolution. Through these innovations, the paper aims to improve the reconstruction quality of dynamic MRI images, particularly in terms of noise reduction and detail preservation, reaching a level comparable to supervised learning methods.

Zero-shot Dynamic MRI Reconstruction with Global-to-local Diffusion Model

Multifold Acceleration of Diffusion MRI Via Deep Learning Reconstruction from Slice-Undersampled Data.

High-Frequency Space Diffusion Models for Accelerated MRI

High-Frequency Space Diffusion Model for Accelerated MRI

Reconstructing High-Quality Diffusion MRI Data from Orthogonal Slice-Undersampled Data Using Graph Convolutional Neural Networks

Learning to reconstruct accelerated MRI through K-space cold diffusion without noise

A Novel Reconstruction Approach Combining Global and Local Low-rank Constraints for Undersampled Dynamic MRI

Universal generative modeling in dual domains for dynamic MRI

K-space Cold Diffusion: Learning to Reconstruct Accelerated MRI without Noise

Undersampled Dynamic MRI Reconstruction by Double Sparse Spatiotemporal Dictionary.

Sub-DM:Subspace Diffusion Model with Orthogonal Decomposition for MRI Reconstruction

CDiffMR: Can We Replace the Gaussian Noise with K-Space Undersampling for Fast MRI?

Correlated and Multi-frequency Diffusion Modeling for Highly Under-sampled MRI Reconstruction

Diffusion Modeling with Domain-conditioned Prior Guidance for Accelerated MRI and qMRI Reconstruction

When Diffusion MRI Meets Diffusion Model: A Novel Deep Generative Model for Diffusion MRI Generation

Diffusion model based on generalized map for accelerated MRI

Physics-informed Deep Diffusion MRI Reconstruction with Synthetic Data: Break Training Data Bottleneck in Artificial Intelligence

Improved Multi-Shot Diffusion-Weighted MRI with Zero-Shot Self-Supervised Learning Reconstruction

Towards performant and reliable undersampled MR reconstruction via diffusion model sampling

Fast Controllable Diffusion Models for Undersampled MRI Reconstruction

Score-based Diffusion Models With Self-supervised Learning For Accelerated 3D Multi-contrast Cardiac Magnetic Resonance Imaging