Book chapter
Dictionary, Structured Low-Rank, and Manifold Learning-Based Reconstruction
Magnetic Resonance Image Reconstruction - Theory Methods and Applications, pp.249-279
Advances in Magnetic Resonance Technology and Applications, v. 7, Academic Press`
2022
DOI: 10.1016/b978-0-12-822726-8.00020-8
Abstract
This chapter reviews learned image representations for accelerated MRI. Unlike prior approaches that use fixed models for image recovery, recent methods rely on advances in low-dimensional models to adapt the representation to the data. The improved representation translates to quite significant gains in acceleration. We review global low-rank representations, local low-rank methods that approximate groups of signals by subspaces, dictionary learning methods, and smooth manifold-based approaches. All of the schemes learn the representation either from exemplar data, calibration data, or jointly learn the representation and images from undersampled data. We also review applications of these frameworks to static and dynamic imaging applications.
Details
- Title: Subtitle
- Dictionary, Structured Low-Rank, and Manifold Learning-Based Reconstruction
- Creators
- Mathews Jacob - University of IowaSajan Goud Lingala - University of IowaMerry Mani - University of Iowa
- Resource Type
- Book chapter
- Publication Details
- Magnetic Resonance Image Reconstruction - Theory Methods and Applications, pp.249-279
- Publisher
- Academic Press`; Amsterdam
- Series
- Advances in Magnetic Resonance Technology and Applications; v. 7
- DOI
- 10.1016/b978-0-12-822726-8.00020-8
- Language
- English
- Date published
- 2022
- Academic Unit
- Radiology; Iowa Neuroscience Institute; Radiation Oncology; Electrical and Computer Engineering; Roy J. Carver Department of Biomedical Engineering
- Record Identifier
- 9984530559202771
Metrics
1 Record Views