Sampling-Pattern-Agnostic MRI Reconstruction through Adaptive Consistency Enforcement with Diffusion Model

TL;DR

This paper introduces a diffusion model-based MRI reconstruction method that is agnostic to sampling patterns, enabling high-quality image recovery across various under-sampling schemes and contrasts, with strong generalisability and improved performance.

Contribution

The paper presents a novel sampling-pattern-agnostic MRI reconstruction approach using diffusion models and adaptive consistency enforcement, enhancing generalisability across sampling patterns and contrasts.

Findings

Outperforms baseline methods in MRI reconstruction tasks.

Successfully generalizes across different sampling trajectories and contrasts.

Validated on MICCAI 2024 CMRxRecon dataset.

Abstract

Magnetic Resonance Imaging (MRI) is a powerful, non-invasive diagnostic tool; however, its clinical applicability is constrained by prolonged acquisition times. Whilst present deep learning-based approaches have demonstrated potential in expediting MRI processes, these methods usually rely on known sampling patterns and exhibit limited generalisability to novel patterns. In the paper, we propose a sampling-pattern-agnostic MRI reconstruction method via a diffusion model through adaptive consistency enforcement. Our approach effectively reconstructs high-fidelity images with varied under-sampled acquisitions, generalising across contrasts and acceleration factors regardless of sampling trajectories. We train and validate across all contrasts in the MICCAI 2024 Cardiac MRI Reconstruction Challenge (CMRxRecon) dataset for the ``Random sampling CMR reconstruction'' task. Evaluation results indicate that our proposed method significantly outperforms baseline methods.