Accelerated Free-Breathing Phase Contrast Cardiovascular MRI using Spatial-Temporal Dictionary Learning Neural

The study presents a novel phase-contrast MRI (PC-MRI) reconstruction framework, employing spatial-temporal dictionary learning neural networks (stDLNN), to tackle the issues of artifacts and noise in undersampled radial acquisitions of dynamic PC images. By integrating dictionary learning and deep learning, the proposed method aims to improve reconstruction accuracy. Comparative results reveal the stDLNN-based approach outperforms existing methods in velocity-encoded image reconstruction and velocity map quality, even at high acceleration rates. This advancement in PC-MRI reconstruction holds promise for real-time imaging, with potential implications for enhancing clinical applications and advancing PC-MRI reconstruction techniques.