Cartesian & Non-Cartesian Sampling Schemes - Advantages & Disadvantages
Maria Altbach1

1University of Arizona, United States

Synopsis

Cartesian sampling is widely used in conventional MRI, however, non-Cartesian sampling schemes (e.g. radial or spiral sampling) offer advantages over Cartesian schemes. Among them is flexibility and efficiency of k-space sampling, motion insensitivity, and the ability to generate images with high spatio-temporal resolution from limited data. The lecture will cover the basic acquisition schemes of Cartesian and non-Cartesian sampling along with the conventional and state-of-the-art reconstruction methods with an emphasis on advantages and disadvantages.

Abstract:

Cartesian sampling is widely used in conventional MRI acquisitions and its popularity has been based on the reconstruction speed afforded by the Fast Fourier Transform (FFT) algorithm. Non-Cartesian sampling schemes (e.g. radial or spiral sampling) offer advantages over Cartesian schemes, among them is flexibility and efficiency of k-space sampling. A key advantage of schemes that oversample the center of k-space is the intrinsic insensitivity to motion and, as shown more recently, the ability to generate images with high spatio-temporal resolution from limited data. The lecture will cover the basic acquisition schemes of Cartesian and Non-Cartesian sampling along with the conventional and state-of-the-art reconstruction methods with an emphasis of their advantages and disadvantages.

Target Audience:

Basic scientist and clinicians who want to better understand the basic principles and practical issues of k-space sampling and reconstruction in terms of the advantages and disadvantages for research and clinical applications.

Introduction:

This section will cover the basics of k-space and sampling theory including the basic reconstruction schemes such as FFT and Non Uniform Fourier Transform (NUFT).

Cartesian acquisition strategies:

This section will cover the full Cartesian and partial Fourier acquisition strategies, highlighting differences in terms of acquisition efficiency as well as the benefits and limitations of partial Fourier. It will cover Echo Planar Imaging (EPI) and the consequences of imperfect Cartesian trajectories.

Non-Cartesian acquisition strategies:

This section will cover 2D and 3D radial and spiral sampling acquisition schemes with an emphasis on advantages and disadvantages of non-uniform k-space sampling. Among the advantages the section will cover the use of these trajectories in terms of motion insensitivity, efficient k-space coverage, high spatio-temporal sampling including view ordering and its implications in parametric imaging. Among the drawbacks of non-Cartesian trajectories, it will cover artifacts such as off-resonance sensitivity, effect of trajectory imperfections, and differences in aliasing artifacts relative to the Cartesian counterpart.

Hybrid trajectories:

This section will cover trajectories that mix Cartesian and non-Cartesian schemes either by sampling on the Cartesian grid but with a non-Cartesian trajectory or by employing non-Cartesian sampling in kx-ky and Cartesian sampling along kz (e.g. radial and spiral stack of stars as well trajectories with pseudo random k-space coverage like DISCO and TWIST).

Reconstruction of non-Cartesian data:

Non-Cartesian acquisitions are best suited to extract unique information from highly undersampled data. This section will provide an overview of novel reconstruction algorithms for undersampled data including the use of multi-coil information, model based reconstruction methods for parametric imaging, and Magnetic Resonance Fingerprinting (MRF). The section will address status on the clinical implementation of these more computationally demanding methods.

Clinical Impact of non-Cartesian sampling:

This section will highlight applications where non-Cartesian methods are preferred due to their intrinsic motion robustness. It will also focus on applications where images with high spatio-temporal resolution are acquired at accelerations not available with conventional Cartesian methods. These include free breathing acquisitions, rapid parameter mapping (e.g., T1/T2 mapping), dynamic contrast enhancement MRI, and other dynamic processes.

The ideal scanner for optimal non-Cartesian scanning:

This section will address limitations of current scanners for optimal non-Cartesian scanning and provide an insight on what changes in hardware and software are needed to take full advantage of non-Cartesian MRI in the clinic.

Acknowledgements

No acknowledgement found.

References

No reference found.
Proc. Intl. Soc. Mag. Reson. Med. 26 (2018)