ISMRM22 - Image Reconstruction: Theory, Methods & Practical Considerations

Joint Annual Meeting ISMRM-ESMRMB & ISMRT 31st Annual Meeting • 07-12 May 2022 • London, UK

Image Reconstruction: Theory, Methods & Practical Considerations

Weekend Course

ORGANIZERS: Fang Liu, Justin Haldar, Dong Liang

Sunday, 08 May 2022

N11 (Breakout A)

07:45 - 11:45

Moderators:
Basics of MRI & Parallel Imaging: Anagha Deshmane
Sparsity & Low Rank: David Atkinson
Practical & Clinical Aspects: Maxim Zaitsev

Skill Level: Basic to Advanced

Session Number: WE-19

Session Number: WE-19

Overview
This course will give an overview of the different techniques for MR image reconstruction, including a basic introduction to k-space and the formulation of image reconstruction as an inverse problem, descriptions of several advanced reconstruction approaches, and practical discussions of software implementation and image quality evaluation.

Target Audience
Anyone who wishes to acquire an understanding of advanced MRI reconstruction techniques.

Educational Objectives
As a result of attending this course, participants should be able to:
- Explain the basic principles of MR image reconstruction;
- Describe the principles and practical significance of several advanced image reconstruction methods; and
- Recognize the practical issues associated with MRI image reconstruction, including practical implementation and image quality evaluation.

	Basics of MRI & Parallel Imaging
07:45		Basics of MRI Reconstruction: k-Space & Inverse Problems View the Presentation Maria Engel In this talk, we look at inverse problems in MR image reconstruction and how to tackle them. We start off with the spatial frequency concept (k-space), which is pivotal for acquisition and reconstruction. We look at different ways of traversing k-space in 2D and 3D and their pros and cons for various applications. We review hardware limitations and ways to account for different k-space trajectories at the image reconstruction stage. We will also investigate the handling of static and dynamic magnetic field inhomogeneities in image reconstruction.
08:10		Parallel Imaging & Simultaneous Multislice Reconstruction View the Presentation Julia Velikina The objective of this presentation is to provide an overview of parallel MR imaging methods and their applications in clinical practice. We will start with discussing image formation for reduced data acquisition and ways to compensate for the missing data with parallel MRI techniques. We will discuss limitations of parallel MRI such as noise amplification and sensitivity to calibration and their effect on achievable acceleration and image artifacts. We will also consider simultaneous multi-slice imaging and discuss its sampling and reconstruction strategies. We will finish with review of some clinical applications that can benefit from the use of parallel imaging.
	Sparsity & Low Rank
08:35		Sparsity-Based Reconstruction View the Presentation Li Feng This talk will present an overview of compressed sensing and its application in rapid MRI.
09:00		Low-Rank & Structured Low-Rank Reconstruction Approaches View the Presentation Bo Zhao
09:25		Break & Meet the Teachers
	Manifold & Deep Learning
09:50		Manifold-Based Reconstruction View the Presentation Sajan Goud Lingala
10:15		Deep Learning-Based Reconstruction View the Presentation Dosik Hwang This talk explains the concept of the deep learning-based MR image reconstruction, from the basics to the up-to-date methods. It covers image-domain deep learning, k-space-domain deep learning, cross-domain deep learning, and direct mapping. The pros and cons of each approaches are explained. Parallel imaging and parameter mapping are also included in the line of deep learning-based MR image reconstruction.
	Practical & Clinical Aspects
10:40		Practical Reconstruction Implementation View the Presentation Zhengguo Tan In this talk, I will start with a brief introduction to an open-source image reconstruction framework. Based on this, I will then present how to implement state-of-the-art compressed sensing regularizers, e.g. total variation and locally low rank (LLR). In particular, I will demonstrate how LLR is solved in the case of multi-contrast acquisition.
11:05		Judgment of Reconstruction Quality View the Presentation Tim Leiner