Abstract

Regularization is a common technique used to improve image quality in inverse problems such as MR image reconstruction. In this work, we extend our previous Graphics Processing Unit (GPU) implementation of MR image reconstruction with compensation for susceptibility-induced field inhomogeneity effects by incorporating an additional quadratic regularization term. Regularization techniques commonly impose the prior information that MR images are relatively smooth by penalizing large changes in intensity between neighboring voxels. However, the associated computations often increase data access and the overall computational load, which can lead to slower image reconstruction. This motivates us to adopt a GPU-enabled implementation of spatial regularization using sparse matrices. This implementation enables the computations for the entire reconstruction procedure to be done on the GPU, which avoids the memory bandwidth bottlenecks associated with frequent communications between the GPU and CPU. Both the CPU and GPU code of this implementation will be available for release at the time of the conference.

Original languageEnglish (US)
Title of host publicationProceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010
Pages578-582
Number of pages5
DOIs
StatePublished - Dec 1 2010
Event3rd International Conference on BioMedical Engineering and Informatics, BMEI 2010 - Yantai, China
Duration: Oct 16 2010Oct 18 2010

Publication series

NameProceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010
Volume2

Other

Other3rd International Conference on BioMedical Engineering and Informatics, BMEI 2010
CountryChina
CityYantai
Period10/16/1010/18/10

Fingerprint

Computer-Assisted Image Processing
Magnetic resonance imaging
Image reconstruction
Program processors
Communication
Inverse problems
Image quality
Graphics processing unit
Bandwidth
Data storage equipment

Keywords

  • CUDA
  • Graphics processing unit (GPU)
  • Iterative reconstruction
  • Magnetic resonance imaging (MRI)
  • Regularization
  • Sparse matrix

ASJC Scopus subject areas

  • Biomedical Engineering
  • Health Informatics
  • Health Information Management

Cite this

Zhuo, Y., Sutton, B., Wu, X. L., Haldar, J., Hwu, W. M., & Liang, Z. P. (2010). Sparse regularization in MRI iterative reconstruction using GPUs. In Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010 (pp. 578-582). [5640008] (Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010; Vol. 2). https://doi.org/10.1109/BMEI.2010.5640008

Sparse regularization in MRI iterative reconstruction using GPUs. / Zhuo, Yue; Sutton, Bradley; Wu, Xiao Long; Haldar, Justin; Hwu, Wen Mei; Liang, Zhi Pei.

Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010. 2010. p. 578-582 5640008 (Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010; Vol. 2).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Zhuo, Y, Sutton, B, Wu, XL, Haldar, J, Hwu, WM & Liang, ZP 2010, Sparse regularization in MRI iterative reconstruction using GPUs. in Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010., 5640008, Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010, vol. 2, pp. 578-582, 3rd International Conference on BioMedical Engineering and Informatics, BMEI 2010, Yantai, China, 10/16/10. https://doi.org/10.1109/BMEI.2010.5640008
Zhuo Y, Sutton B, Wu XL, Haldar J, Hwu WM, Liang ZP. Sparse regularization in MRI iterative reconstruction using GPUs. In Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010. 2010. p. 578-582. 5640008. (Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010). https://doi.org/10.1109/BMEI.2010.5640008
Zhuo, Yue ; Sutton, Bradley ; Wu, Xiao Long ; Haldar, Justin ; Hwu, Wen Mei ; Liang, Zhi Pei. / Sparse regularization in MRI iterative reconstruction using GPUs. Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010. 2010. pp. 578-582 (Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010).
@inproceedings{51b65e3c88c4411daf5f6be3d31a62dc,
title = "Sparse regularization in MRI iterative reconstruction using GPUs",
abstract = "Regularization is a common technique used to improve image quality in inverse problems such as MR image reconstruction. In this work, we extend our previous Graphics Processing Unit (GPU) implementation of MR image reconstruction with compensation for susceptibility-induced field inhomogeneity effects by incorporating an additional quadratic regularization term. Regularization techniques commonly impose the prior information that MR images are relatively smooth by penalizing large changes in intensity between neighboring voxels. However, the associated computations often increase data access and the overall computational load, which can lead to slower image reconstruction. This motivates us to adopt a GPU-enabled implementation of spatial regularization using sparse matrices. This implementation enables the computations for the entire reconstruction procedure to be done on the GPU, which avoids the memory bandwidth bottlenecks associated with frequent communications between the GPU and CPU. Both the CPU and GPU code of this implementation will be available for release at the time of the conference.",
keywords = "CUDA, Graphics processing unit (GPU), Iterative reconstruction, Magnetic resonance imaging (MRI), Regularization, Sparse matrix",
author = "Yue Zhuo and Bradley Sutton and Wu, {Xiao Long} and Justin Haldar and Hwu, {Wen Mei} and Liang, {Zhi Pei}",
year = "2010",
month = "12",
day = "1",
doi = "10.1109/BMEI.2010.5640008",
language = "English (US)",
isbn = "9781424464968",
series = "Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010",
pages = "578--582",
booktitle = "Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010",

}

TY - GEN

T1 - Sparse regularization in MRI iterative reconstruction using GPUs

AU - Zhuo, Yue

AU - Sutton, Bradley

AU - Wu, Xiao Long

AU - Haldar, Justin

AU - Hwu, Wen Mei

AU - Liang, Zhi Pei

PY - 2010/12/1

Y1 - 2010/12/1

N2 - Regularization is a common technique used to improve image quality in inverse problems such as MR image reconstruction. In this work, we extend our previous Graphics Processing Unit (GPU) implementation of MR image reconstruction with compensation for susceptibility-induced field inhomogeneity effects by incorporating an additional quadratic regularization term. Regularization techniques commonly impose the prior information that MR images are relatively smooth by penalizing large changes in intensity between neighboring voxels. However, the associated computations often increase data access and the overall computational load, which can lead to slower image reconstruction. This motivates us to adopt a GPU-enabled implementation of spatial regularization using sparse matrices. This implementation enables the computations for the entire reconstruction procedure to be done on the GPU, which avoids the memory bandwidth bottlenecks associated with frequent communications between the GPU and CPU. Both the CPU and GPU code of this implementation will be available for release at the time of the conference.

AB - Regularization is a common technique used to improve image quality in inverse problems such as MR image reconstruction. In this work, we extend our previous Graphics Processing Unit (GPU) implementation of MR image reconstruction with compensation for susceptibility-induced field inhomogeneity effects by incorporating an additional quadratic regularization term. Regularization techniques commonly impose the prior information that MR images are relatively smooth by penalizing large changes in intensity between neighboring voxels. However, the associated computations often increase data access and the overall computational load, which can lead to slower image reconstruction. This motivates us to adopt a GPU-enabled implementation of spatial regularization using sparse matrices. This implementation enables the computations for the entire reconstruction procedure to be done on the GPU, which avoids the memory bandwidth bottlenecks associated with frequent communications between the GPU and CPU. Both the CPU and GPU code of this implementation will be available for release at the time of the conference.

KW - CUDA

KW - Graphics processing unit (GPU)

KW - Iterative reconstruction

KW - Magnetic resonance imaging (MRI)

KW - Regularization

KW - Sparse matrix

UR - http://www.scopus.com/inward/record.url?scp=78650661692&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78650661692&partnerID=8YFLogxK

U2 - 10.1109/BMEI.2010.5640008

DO - 10.1109/BMEI.2010.5640008

M3 - Conference contribution

AN - SCOPUS:78650661692

SN - 9781424464968

T3 - Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010

SP - 578

EP - 582

BT - Proceedings - 2010 3rd International Conference on Biomedical Engineering and Informatics, BMEI 2010

ER -