From Compressed-Sensing to Artificial Intelligence-Based Cardiac MRI Reconstruction

Abstract

Cardiac magnetic resonance (CMR) imaging is an important tool for the non-invasive assessment of cardiovascular disease. However, CMR suffers from long acquisition times due to the need of obtaining images with high temporal and spatial resolution, different contrasts, and/or whole-heart coverage. In addition, both cardiac and respiratory-induced motion of the heart during the acquisition need to be accounted for, further increasing the scan time. Several undersampling reconstruction techniques have been proposed during the last decades to speed up CMR acquisition. These techniques rely on acquiring less data than needed and estimating the non-acquired data exploiting some sort of prior information. Parallel imaging and compressed sensing undersampling reconstruction techniques have revolutionized the field, enabling 2-to 3-fold scan time accelerations to become standard in clinical practice. Recent scientific advances in CMR reconstruction hinge on the thriving field of artificial intelligence. Machine learning reconstruction approaches have been recently proposed to learn the non-linear optimization process employed in CMR reconstruction. Unlike analytical methods for which the reconstruction problem is explicitly defined into the optimization process, machine learning techniques make use of large data sets to learn the key reconstruction parameters and priors. In particular, deep learning techniques promise to use deep neural networks (DNN) to learn the reconstruction process from existing datasets in advance, providing a fast and efficient reconstruction that can be applied to all newly acquired data. However, before machine learning and DNN can realize their full potentials and enter widespread clinical routine for CMR image reconstruction, there are several technical hurdles that need to be addressed. In this article, we provide an overview of the recent developments in the area of artificial intelligence for CMR image reconstruction. The underlying assumptions of established techniques such as compressed sensing and low-rank reconstruction are briefly summarized, while a greater focus is given to recent advances in dictionary learning and deep learning based CMR reconstruction. In particular, approaches that exploit neural networks as implicit or explicit priors are discussed for 2D dynamic cardiac imaging and 3D whole-heart CMR imaging. Current limitations, challenges, and potential future directions of these techniques are also discussed.

Role of Cardiac Magnetic Resonance to Improve Risk Prediction following Acute ST-elevation Myocardial Infarction

“…These results could be corroborated by Bulluck et al [68]; however, these small studies could not validate their findings for hard clinical events, which remains to be done by future studies. Compressed sensing has not only been investigated for accelerating cine CMR imaging but also shortening scan time of LGE imaging [69]. Moreover, data on an accelerated, free-breathing 3D T1 mapping technique for contrast-free myocardial tissue characterization have been published recently [70].…”

“…Despite the promising data on compressed sensing, there remain relevant drawbacks with this technique (e.g., robustness and time of the reconstruction). Whether machine learning techniques can provide a solution for a fast and efficient reconstruction of newly acquired data is intensively studied now [69].…”

“…Simple, fast, and robust protocols allowing high quality CMR infarct characterization are required to make CMR more feasible in daily clinical routine. Although there have been progresses in accelerating the acquisition of CMR, validation and prognostic evaluation of such protocols are currently lacking [3,11,69,158]. Even more important, there are thus far no data from prospective randomized controlled trials showing that CMR-guided risk stratification improves patient outcome.…”

Speeding Up Ultrafast Spectroscopy

“…As a mathematical algorithm for fitting and reconstructing experimental data, CS falls into this latter category. Developed in the 1990s and 2000s [2][3][4], the approach is often associated with the fields of nuclear magnetic resonance (NMR) and spatial imaging, where it was first deployed [4][5][6][7]. But the concept can be applied to virtually any experimental signal.…”

“…Obtaining such measurements using a conventional signal-processing method would usually involve scanning over the full parameter space in every variable. By downsampling this parameter space for each variable, CS allows these measurements to be made with significant data-acquisition-rate benefits [1,[4][5][6][7][8]. This efficiency improvement has the added advantage that samples are less prone to degradation from long exposure to the probe.…”