Synopsis
This presentation aims to achieve three tasks: (a) outline recent developments in non-Cartesian cardiac MRI approaches used for myocardial tissue characterization and myocardial perfusion imaging; (b) describe continuous data acquisition techniques which aim to capture the dynamics of the function/anatomy of the heart without the need for breath-holding and/or ECG gating; and (c) provide an overview of the evolving approaches for joint parametric mapping techniques including myocardial MR fingerprinting and multi-tasking which aim to drastically simplify the workflow in clinical cardiac MR protocols.
Overview
The diagnostic and prognostic utility of cardiac MR (CMR) imaging has continued to evolve with the development of new techniques for the detection and characterization of heart disease including providing unprecedented tissue characterization of the myocardium. Although well-known markers of heart disease, such as perfusion and viability, using MRI have been successfully validated against nuclear methods in multi-center studies, and are routinely used in the clinical area, recent advances in cardiac MRI acquisition strategies and pulse sequences are pointing to exciting new directions.
Novel acquisition strategies combined with advanced reconstruction techniques are enabling high-resolution, truly 3D dynamic acquisitions which will be the future of CMR. Furthermore, techniques such as MR fingerprinting and MR multi-tasking may greatly facilitate the more widespread clinical adoption of CMR, by making CMR more user-friendly. In this presentation we will focus on three main areas: (a) recent developments in non-Cartesian acquisition strategies; (b) continuous data-acquisition techniques; (c) joint parametric mapping techniques including myocardial MR fingerprinting and MR multi-tasking.
Recent developments in non-Cartesian acquisition strategies
We will discuss the inherent advantages that non-Cartesian sampling techniques (most commonly, radial and spiral trajectories) provide for cardiovascular imaging including their ability to reduce the severity and prevalence of image artifacts due to their robustness to motion. We will review recently published work involving pulse sequences that employ radial and spiral k-space sampling for myocardial perfusion imaging and myocardial tissue characterization. Unlike nuclear imaging methods which intrinsically achieve whole-heart coverage, extending the spatial coverage of CMR-based myocardial characterization methods (T1 mapping, T2 mapping, perfusion quantification) to whole-heart involves technical trade-offs including a somewhat lower spatial resolution and a larger temporal acquisition window (in each cardiac cycle), which could lead to a higher prevalence of image artifacts. Overall, 3D myocardial CMR involves a high level of computational complexity, hence limiting its current availability to specialized research centers. Nevertheless, whole-heart myocardial CMR techniques are being extensively pursued to address these technical challenges. Continuous data acquisition techniques
This relatively new class of CMR pulse sequences aims to capture the dynamics of the function/anatomy of the heart using continuous data acquisition without the need for breath-holding and/or ECG triggering. Specifically, recent work has demonstrated the feasibility of high-resolution perfusion CMR without the need for ECG gating. Such data acquisition methods use a continuous golden-angle radial sampling method (with nearly 10-fold acceleration) where the ECG signal is ignored and instead the timing of acquired dynamics is detected retrospectively form a “navigator” image. In the presence of breathing motion, an emerging approach in is to deal with the respiratory motion by resolving it (or imaging it as a new dimension of the underlying dynamic object) instead of trying to perform motion correction/compensation during or after the image reconstruction process. Instead of discarding data or enforcing motion models for motion correction, this new class of continuous imaging methods make constructive use of all respiratory phases to generate a multi-dimensional respiratory-resolved view of the heart. Joint parametric mapping techniques
These approaches have received significant attention in the field and include myocardial MR fingerprinting and MR multi-tasking. The principle behind these approaches is to continuously acquire data with a non-Cartesian pulse sequence, e.g., using radial golden-angle sampling or spiral sampling, which may allow drastic simplification of the workflow or “push-button” CMR and consequently facilitate wider clinical adoption of CMR. One of the main goals is to improve the workflow limitations in the clinical assessment of heart disease by enabling concurrent estimation of myocardial characteristics (T1, T2, ECV) and in some cases also cardiac function and perfusion during a single scan. We will provide a brief overview of MR fingerprinting and MR multi-tasking and will describe the ongoing work in improving the robustness of each approach for myocardial T1 and T2 mapping. Future directions
In the final part of the presentation, we will touch on an emerging area related to the impact that artificial intelligence (deep learning methods) is poised to have on CMR pulse sequences within the time horizon of the next five years. Specifically, we will discuss how rapid deep-learning-based image reconstruction has the potential to enable “patient adaptive” data-acquisition optimization by providing “on the fly” feedback from the reconstructed images to the CMR pulse sequence. We will also discuss the need to provide rigorous spatio-temporal resolution targets for a comprehensive range of CMR diagnostics tasks to guide the future development of CMR pulse sequences. Acknowledgements
No acknowledgement found.References
No reference found.