Free-running time-resolved first-pass myocardial perfusion using a multi-scale dynamics decomposition: CMR-MOTUS

Objective: First-pass myocardial perfusion involves several types of dynamics, including cardiac motion, respiratory motion, bulk motion and contrast agent inflow. To accurately quantify the initial inflow of the contrast agent, high spatiotemporal resolution MR imaging must be obtained. To achieve this, we present a novel approach, named CMR-MOTUS, for the reconstruction of time-resolved free-running first-pass myocardial perfusion by jointly estimating high-quality motion fields and contrast-varying images. Materials and methods: We propose CMR-MOTUS, which extends the MR-MOTUS framework by integrating a contrast-varying reference image with a low-rank plus sparse decomposition to capture additional dynamics such as blood flow and contrast agent inflow. This joint reconstruction framework alternates between solving for time-dependent image contrast changes and motion fields, eliminating the need for a pre-acquisition motion-static reference image. The method was tested on simulations and in-vivo datasets. Results: In simulations, CMR-MOTUS showed improved image similarity and motion field accuracy compared to state-of-the-art methods. In in-vivo tests, the methods effectively captured cardiac and respiratory motion dynamics, resulting in cine images with sharper features than state-of-the-art. Discussion: CMR-MOTUS presents significant advantages by modelling motion and contrast dynamics in the reconstruction of first-pass myocardial perfusion. The framework enables a data-efficient free-running workflow since the entire acquisition is correlated with high-quality motion fields. This approach has the potential to enhance the diagnostic value of cardiac MRI but needs further clinical validations.