More IMPATIENT: A gridding-accelerated Toeplitz-based strategy for non-Cartesian high-resolution 3D MRI on GPUs

Jiading Gai, Nady Obeid, Joseph L. Holtrop, Xiao Long Wu, Fan Lam, Maojing Fu, Justin P. Haldar, Wen Mei W Hwu, Zhi Pei Liang, Bradley P. Sutton

Research output: Contribution to journalArticle

Abstract

Several recent methods have been proposed to obtain significant speed-ups in MRI image reconstruction by leveraging the computational power of GPUs. Previously, we implemented a GPU-based image reconstruction technique called the Illinois Massively Parallel Acquisition Toolkit for Image reconstruction with ENhanced Throughput in MRI (IMPATIENT MRI) for reconstructing data collected along arbitrary 3D trajectories. In this paper, we improve IMPATIENT by removing computational bottlenecks by using a gridding approach to accelerate the computation of various data structures needed by the previous routine. Further, we enhance the routine with capabilities for off-resonance correction and multi-sensor parallel imaging reconstruction. Through implementation of optimized gridding into our iterative reconstruction scheme, speed-ups of more than a factor of 200 are provided in the improved GPU implementation compared to the previous accelerated GPU code.

Original languageEnglish (US)
Pages (from-to)686-697
Number of pages12
JournalJournal of Parallel and Distributed Computing
Volume73
Issue number5
DOIs
StatePublished - May 2013

Fingerprint

Otto Toeplitz
Image Reconstruction
Magnetic resonance imaging
High Resolution
Image reconstruction
Accelerate
Data Structures
Throughput
Imaging
Trajectory
Data structures
Trajectories
Arbitrary
Imaging techniques
Strategy
Graphics processing unit
Sensors

Keywords

  • CUDA
  • GPU
  • Gridding
  • MRI
  • Non-Cartesian
  • Toeplitz

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Networks and Communications
  • Hardware and Architecture
  • Software
  • Theoretical Computer Science

Cite this

More IMPATIENT : A gridding-accelerated Toeplitz-based strategy for non-Cartesian high-resolution 3D MRI on GPUs. / Gai, Jiading; Obeid, Nady; Holtrop, Joseph L.; Wu, Xiao Long; Lam, Fan; Fu, Maojing; Haldar, Justin P.; Hwu, Wen Mei W; Liang, Zhi Pei; Sutton, Bradley P.

In: Journal of Parallel and Distributed Computing, Vol. 73, No. 5, 05.2013, p. 686-697.

Research output: Contribution to journalArticle

Gai, Jiading ; Obeid, Nady ; Holtrop, Joseph L. ; Wu, Xiao Long ; Lam, Fan ; Fu, Maojing ; Haldar, Justin P. ; Hwu, Wen Mei W ; Liang, Zhi Pei ; Sutton, Bradley P. / More IMPATIENT : A gridding-accelerated Toeplitz-based strategy for non-Cartesian high-resolution 3D MRI on GPUs. In: Journal of Parallel and Distributed Computing. 2013 ; Vol. 73, No. 5. pp. 686-697.
@article{e8b11aa8bb0a408c89e39c03ed65c595,
title = "More IMPATIENT: A gridding-accelerated Toeplitz-based strategy for non-Cartesian high-resolution 3D MRI on GPUs",
abstract = "Several recent methods have been proposed to obtain significant speed-ups in MRI image reconstruction by leveraging the computational power of GPUs. Previously, we implemented a GPU-based image reconstruction technique called the Illinois Massively Parallel Acquisition Toolkit for Image reconstruction with ENhanced Throughput in MRI (IMPATIENT MRI) for reconstructing data collected along arbitrary 3D trajectories. In this paper, we improve IMPATIENT by removing computational bottlenecks by using a gridding approach to accelerate the computation of various data structures needed by the previous routine. Further, we enhance the routine with capabilities for off-resonance correction and multi-sensor parallel imaging reconstruction. Through implementation of optimized gridding into our iterative reconstruction scheme, speed-ups of more than a factor of 200 are provided in the improved GPU implementation compared to the previous accelerated GPU code.",
keywords = "CUDA, GPU, Gridding, MRI, Non-Cartesian, Toeplitz",
author = "Jiading Gai and Nady Obeid and Holtrop, {Joseph L.} and Wu, {Xiao Long} and Fan Lam and Maojing Fu and Haldar, {Justin P.} and Hwu, {Wen Mei W} and Liang, {Zhi Pei} and Sutton, {Bradley P.}",
year = "2013",
month = "5",
doi = "10.1016/j.jpdc.2013.01.001",
language = "English (US)",
volume = "73",
pages = "686--697",
journal = "Journal of Parallel and Distributed Computing",
issn = "0743-7315",
publisher = "Academic Press Inc.",
number = "5",

}

TY - JOUR

T1 - More IMPATIENT

T2 - A gridding-accelerated Toeplitz-based strategy for non-Cartesian high-resolution 3D MRI on GPUs

AU - Gai, Jiading

AU - Obeid, Nady

AU - Holtrop, Joseph L.

AU - Wu, Xiao Long

AU - Lam, Fan

AU - Fu, Maojing

AU - Haldar, Justin P.

AU - Hwu, Wen Mei W

AU - Liang, Zhi Pei

AU - Sutton, Bradley P.

PY - 2013/5

Y1 - 2013/5

N2 - Several recent methods have been proposed to obtain significant speed-ups in MRI image reconstruction by leveraging the computational power of GPUs. Previously, we implemented a GPU-based image reconstruction technique called the Illinois Massively Parallel Acquisition Toolkit for Image reconstruction with ENhanced Throughput in MRI (IMPATIENT MRI) for reconstructing data collected along arbitrary 3D trajectories. In this paper, we improve IMPATIENT by removing computational bottlenecks by using a gridding approach to accelerate the computation of various data structures needed by the previous routine. Further, we enhance the routine with capabilities for off-resonance correction and multi-sensor parallel imaging reconstruction. Through implementation of optimized gridding into our iterative reconstruction scheme, speed-ups of more than a factor of 200 are provided in the improved GPU implementation compared to the previous accelerated GPU code.

AB - Several recent methods have been proposed to obtain significant speed-ups in MRI image reconstruction by leveraging the computational power of GPUs. Previously, we implemented a GPU-based image reconstruction technique called the Illinois Massively Parallel Acquisition Toolkit for Image reconstruction with ENhanced Throughput in MRI (IMPATIENT MRI) for reconstructing data collected along arbitrary 3D trajectories. In this paper, we improve IMPATIENT by removing computational bottlenecks by using a gridding approach to accelerate the computation of various data structures needed by the previous routine. Further, we enhance the routine with capabilities for off-resonance correction and multi-sensor parallel imaging reconstruction. Through implementation of optimized gridding into our iterative reconstruction scheme, speed-ups of more than a factor of 200 are provided in the improved GPU implementation compared to the previous accelerated GPU code.

KW - CUDA

KW - GPU

KW - Gridding

KW - MRI

KW - Non-Cartesian

KW - Toeplitz

UR - http://www.scopus.com/inward/record.url?scp=84893661530&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84893661530&partnerID=8YFLogxK

U2 - 10.1016/j.jpdc.2013.01.001

DO - 10.1016/j.jpdc.2013.01.001

M3 - Article

AN - SCOPUS:84893661530

VL - 73

SP - 686

EP - 697

JO - Journal of Parallel and Distributed Computing

JF - Journal of Parallel and Distributed Computing

SN - 0743-7315

IS - 5

ER -