Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

HeinzelCluster: accelerated reconstruction for FORE and OSEM3D.

S Vollmar1, C Michel, J T Treffert

  • 1Max-Planck-Institut für neurologische Forschung, Gleueler Str. 50, 50931 Köln, Germany. vollmar@pet.mpin-koeln.mpg.de

Physics in Medicine and Biology
|August 31, 2002
PubMed
Summary

Accelerating positron emission tomography (PET) reconstruction on high-resolution research tomographs (HRRT) is achieved through parallel processing. Two methods, 2D sinogram transformation and OSEM3D parallelization, significantly reduce reconstruction times.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

[Rare diseases recognizable from blood smears].

Der Internist·2018
Same author

[Angiokeratoma circumscriptum neviforme].

Annales de dermatologie et de venereologie·2018
Same author

[Impact of neoadjuvant chemotherapy on the peri-operative morbidity of radical cystectomy for muscle invasive bladder cancer].

Progres en urologie : journal de l'Association francaise d'urologie et de la Societe francaise d'urologie·2018
Same author

Search for Tensor, Vector, and Scalar Polarizations in the Stochastic Gravitational-Wave Background.

Physical review letters·2018
Same author

Prospects for observing and localizing gravitational-wave transients with Advanced LIGO, Advanced Virgo and KAGRA.

Living reviews in relativity·2018
Same author

GW170817: Implications for the Stochastic Gravitational-Wave Background from Compact Binary Coalescences.

Physical review letters·2018

Area of Science:

  • Medical Imaging
  • Computational Science

Background:

  • Iterative 3D reconstruction for high-resolution research tomograph (HRRT) positron emission tomography (PET) is computationally intensive.
  • Single-processor machines struggle with the excessive computing time required for large sinogram sizes.

Purpose of the Study:

  • To develop and evaluate methods for accelerating 3D PET reconstruction on the HRRT.
  • To overcome computational limitations and enable faster image reconstruction.

Main Methods:

  • Transforming 3D PET scans into 2D sinograms for parallel reconstruction using Fourier rebinning (FORE) and fast 2D methods.
  • Parallelizing the OSEM3D algorithm with multi-level (fine-grain and coarse-grain) parallelization on a dedicated cluster.
  • Comparing reconstruction performance across different communication methods (RPC, Syngo, Myrinet GM).

Related Experiment Videos

Main Results:

  • Achieved a speedup factor greater than 23 using 26 processors with the 2D sinogram transformation approach.
  • Reduced OSEM3D reconstruction time for one core iteration from over 7 hours to approximately 35 minutes through parallelization.
  • Demonstrated the feasibility of parallel processing for significantly reducing PET image reconstruction times.

Conclusions:

  • Parallel processing strategies, including 2D sinogram transformation and OSEM3D parallelization, are effective for accelerating HRRT PET reconstruction.
  • These optimized methods make advanced 3D reconstruction techniques more feasible on available hardware.
  • The study highlights the importance of computational optimization in high-resolution medical imaging.