Motion-Guided Deep Image Prior for Cardiac MRI

2024-12-05

arXiv.org

AI Summary

Cardiac MRI in patients with arrhythmia or breath-holding difficulty suffers from severe motion artifacts due to inconsistent respiratory compliance, hindering reliable dynamic imaging. Method: We propose an unsupervised Motion-Guided Deep Image Prior (MG-DIP) reconstruction framework enabling real-time, free-breathing dynamic cardiac MRI. For the first time, MG-DIP jointly models a time-varying non-rigid deformation field and a spatial dictionary of anatomical templates to decouple physiological motion (cardiac contraction and respiration) from inter-frame content variation. It integrates dictionary-driven synthesis, temporal deformation modeling, and DIP-based optimization—requiring no paired training data or motion ground truth. Results: Evaluated on MRXCAT simulations and clinical free-breathing cine and late gadolinium enhancement (LGE) datasets, MG-DIP consistently outperforms state-of-the-art supervised and unsupervised methods, achieving significant PSNR/SSIM improvements. Radiologist blind assessments further confirm superior diagnostic quality and clinical utility.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Deep Generative Models & Autoencoders
• Intelligent Robots: Motion and Path Planning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Cardiovascular magnetic resonance imaging is a powerful diagnostic tool for assessing cardiac structure and function. Traditional breath-held imaging protocols, however, pose challenges for patients with arrhythmias or limited breath-holding capacity. We introduce Motion-Guided Deep Image prior (M-DIP), a novel unsupervised reconstruction framework for accelerated real-time cardiac MRI. M-DIP employs a spatial dictionary to synthesize a time-dependent template image, which is further refined using time-dependent deformation fields that model cardiac and respiratory motion. Unlike prior DIP-based methods, M-DIP simultaneously captures physiological motion and frame-to-frame content variations, making it applicable to a wide range of dynamic applications. We validate M-DIP using simulated MRXCAT cine phantom data as well as free-breathing real-time cine and single-shot late gadolinium enhancement data from clinical patients. Comparative analyses against state-of-the-art supervised and unsupervised approaches demonstrate M-DIP’s performance and versatility. M-DIP achieved better image quality metrics on phantom data, as well as higher reader scores for in-vivo patient data.

Problem

• Enables high-quality free-breathing cardiac MRI reconstruction
• Captures physiological motion and content variations simultaneously
• Improves image quality without external training data

Innovation

• Unsupervised reconstruction framework for cardiac MRI
• Spatial dictionary synthesizes time-dependent intermediate image
• Time-dependent deformation fields model physiological motion