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

Globally convergent image reconstruction for emission tomography using relaxed ordered subsets algorithms.

Sangtae Ahn1, Jeffrey A Fessler

  • 1Electrical Engineering and Computer Science Department, University of Michigan, 4415 Electrical Engineering and Computer Science Building, 1301 Beal Avenue, Ann Arbor, MI 48109-2122, USA. sangtaea@umich.edu

IEEE Transactions on Medical Imaging
|July 9, 2003
PubMed
Summary

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

Editorial: Deep learning in brain-computer interfaces.

Frontiers in human neuroscience·2026
Same author

A vendor-neutral functional MRI acquisition protocol for multi-site studies.

Aperture neuro·2026
Same author

Phantom- and simulation-based validation of combined diffusion relaxometry in ex vivo ADRD white matter.

bioRxiv : the preprint server for biology·2026
Same author

Smooth optimization using global and local low-rank regularizers.

SIAM journal on imaging sciences·2026
Same author

Bilevel Optimized Implicit Neural Representation for Scan-Specific Accelerated MRI Reconstruction.

IEEE transactions on medical imaging·2026
Same author

Scan-Adaptive MRI Undersampling Using Neighbor-based Optimization (SUNO).

IEEE transactions on computational imaging·2026

We developed new penalized-likelihood algorithms for emission tomography image reconstruction. These modified algorithms ensure global convergence while maintaining fast initial speeds, improving accuracy in medical imaging.

Area of Science:

  • Medical Imaging
  • Computational Science
  • Image Reconstruction

Background:

  • Penalized-likelihood methods are crucial for emission tomography image reconstruction.
  • Existing algorithms like BSREM and OS-SPS have limitations in convergence proofs or behavior.

Purpose of the Study:

  • To develop globally convergent relaxed ordered subsets (OS) algorithms for penalized-likelihood image reconstruction.
  • To address limitations in existing BSREM and OS-SPS algorithms.
  • To improve the reliability and efficiency of image reconstruction in emission tomography.

Main Methods:

  • Modification of block sequential regularized expectation-maximization (BSREM) scaling functions for proven convergence.
  • Introduction of relaxation into ordered subsets separable paraboloidal surrogates (OS-SPS) to prevent limit cycles.

Related Experiment Videos

  • Proof of global convergence for diagonally scaled incremental gradient methods, including relaxed OS-SPS.

    • Main Results:

    • Modified BSREM achieves global convergence under realistic assumptions with convenient stepsize selection.
    • Relaxed OS-SPS algorithm demonstrates global convergence, overcoming limit cycle issues.
    • Both new algorithms exhibit fast initial convergence, comparable to conventional OS methods.

    Conclusions:

    • The presented modified BSREM and relaxed OS-SPS algorithms offer reliable global convergence for penalized-likelihood image reconstruction.
    • These advancements enhance the accuracy and efficiency of emission tomography imaging.
    • The study provides robust algorithmic solutions for challenging image reconstruction problems.