Computational MRI: Beyond Compressed Sensing
- 6 speakers with 30 minute lectures
Speakers: Mariya Doneva - Speaker Website
Philips Research Hamburg
Jeffrey Fessler - Speaker Website
William L. Root Collegiate Professor
Electrical Engineering and Computer Science
The University of Michigan
Sajan Lingala - Speaker Website
Assistant Professor
Roy J Carver Department of Biomedical Engineering
University of Iowa
Jonathan Tamir - Speaker Website
Assistant Professor
Electrical and Computer Engineering
The University of Texas at Austin (UT Austin)
Mathews Jacob - Speaker Website
Professor
Department of Electrical and Computer Engieering
University of Iowa
Mehmet Akcakaya - Speaker Website
McKnight-Land Grant Assistant Professor,
Department of Electrical and Computer Engineering, and Center for Magnetic
Resonance Research
University of Minnesota
Abstracts:
Talk 1: Physics based models for MR image reconstruction
Speaker: Mariya Doneva
The signal measurement model is the link between data acquisition and image reconstruction. Simple models are tempting, and in MRI the Fourier model has been predominantly used because of its simplicity and efficiency. However, this simple model ignores many factors and is often insufficient. This lecture will give an overview of modern techniques for model-based reconstruction that apply an extended signal model to better describe the data acquisition process.
Talk 2: Optimization methods for MR image reconstruction
Speaker: Jeffrey Fessler
This tutorial will summarize several key models and optimization algorithms for MR image reconstruction, including both the type of methods that have been recently FDA approved for clinical use, as well as more recent methods being considered in the research community that use data-adaptive regularizers. Many algorithms have been devised that exploit the structure of the system model and regularizers used in MRI; this talk will collect such algorithms in a single survey. Many of the ideas used in optimization methods for MRI are also useful for solving other inverse problems.
Talk 3: Accelerating dynamic MRI using learned representations
Speaker: Sajan Lingala
In this tutorial, we will offer a unified view of several different approaches to dynamic MRI using learned representations, focusing on the signal processing aspects which make each class of learning methods so powerful. These methods include low-rank methods, blind compressed sensing methods, higher-order multidynamic methods, explicit motion estimation compensated recovery methods, and manifold regularized recovery methods. Over the past five years or so, these schemes have revolutionized dynamic MRI for many applications, offering exciting new capabilities in biomedical imaging. We expect this tutorial to benefit the ISBI audience who are interested in learning on recent trends in dynamic MRI.
Talk 4: Multi-contrast and Quantitative MRI methods
Speaker: Jonathan Tamir
This tutorial will introduce physics-based modeling constraints in MRI and shows how they can be used in conjunction with compressed sensing for image reconstruction and quantitative imaging. We describe model-based quantitative MRI, as well as its linear subspace approximation. We also discuss approaches to selecting user-controllable scan parameters given knowledge of the physical model. We will present MRI applications that take advantage of this framework for the purpose of multi-contrast imaging and quantitative mapping. We expect this tutorial to benefit the ISBI audience who are interested in learning on recent trends in multicontrast and quantitative MRI.
Talk 5: Structured low-rank algorithms for computational MRI
Speaker: Mathews Jacob
We will provide a review of recent advances in the recovery of continuous domain multidimensional signals from their few non-uniform (multichannel) measurements using structured low-rank matrix completion formulation. This framework is centered on the fundamental duality between the compactness (e.g., sparsity) of the continuous signal and the rank of a structured matrix, whose entries are functions of the signal. This property enables the reformulation of MRI signal recovery as a lowrank matrix completion of the structured matrix, which comes with performance guarantees. We will also review fast algorithms that are comparable in complexity to current compressed sensing methods, which enables the application of the framework to large-scale magnetic resonance (MR) recovery problems. The remarkable flexibility of the formulation enables us to exploit signal properties that are difficult to capture by current sparse and low-rank optimization strategies. We will discuss the framework’s utility in a wide range of MR imaging (MRI) applications, including highly accelerated imaging, calibration-free acquisition, MR artifact correction, and ungated dynamic MRI.
Talk 6: Deep learning reconstruction for parallel MRI
Speaker: Mehmet Akcakaya
Parallel imaging is the most commonly used strategy for accelerating MRI acquisitions. These techniques utilize the redundancies in the data acquired by multiple receivers in coil arrays. In this talk, we will overview the recent machine learning approaches that have been proposed specifically for improving parallel imaging. This talk will provide an extensive overview of deep learning methods for parallel MRI. We expect that this survey will be of significant interest to researchers in related fields, and those interested in learning more about improved parallel imaging techniques, building on the clinical standard for accelerated MRI.