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 Video

Updated: Dec 22, 2025

Photorealistic Learned Landscapes for Augmented Reality
06:54

Photorealistic Learned Landscapes for Augmented Reality

Published on: June 27, 2025

589

Generative adversarial network based regularized image reconstruction for PET.

Zhaoheng Xie1, Reheman Baikejiang1, Tiantian Li1

  • 1Department of Biomedical Engineering University of California Davis CA United States of America.

Physics in Medicine and Biology
|May 2, 2020
PubMed
Summary
This summary is machine-generated.

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

Common carp USP30s restrict SVCV replication by attenuating mitophagy.

Fish & shellfish immunology·2026
Same author

Variation in Tier 3 Specialist Weight Management Services Within the National Health Service (NHS): Insights From Four NHS Sites in England and Wales.

Clinical obesity·2026
Same author

Discovery of Novel Alkynylbenzene Scaffold-Based PTPN2-Selective Degrader PD-305 with Exceptional Potency and In Vivo Efficacy.

Journal of medicinal chemistry·2026
Same author

An Au<sub>25</sub> Nanocluster-Based Electrochemical Biosensor for Profenofos Detection With High Activity and Sensitivity.

Chemistry (Weinheim an der Bergstrasse, Germany)·2026
Same author

Biosynthesis of Nitrile-Containing Amino Acids for Rapid Protein Conjugation and Fluorogenic Labeling.

Chem·2026
Same author

Dispersion Mechanism of Conductive Phase Materials and Micro-Mechanical Properties of ERCC.

Materials (Basel, Switzerland)·2026

This study introduces a generative adversarial network (GAN) to enhance Positron Emission Tomography (PET) image reconstruction. The novel method improves lesion contrast and reduces noise, outperforming existing techniques.

Area of Science:

  • Medical Imaging
  • Computational Science
  • Artificial Intelligence

Background:

  • Positron Emission Tomography (PET) image reconstruction is an ill-posed inverse problem.
  • High noise levels in PET images result from limited detected events.
  • Prior information and deep neural networks can improve PET image quality.

Purpose of the Study:

  • To develop an improved PET image reconstruction method using generative adversarial networks (GANs).
  • To enhance the performance of deep neural network-based regularization for PET imaging.
  • To improve lesion contrast recovery while reducing background noise in PET images.

Main Methods:

  • A generative adversarial network (GAN) was proposed for PET image reconstruction.
  • A pretrained denoising neural network was used within a constrained maximum likelihood estimation framework.

Related Experiment Videos

Last Updated: Dec 22, 2025

Photorealistic Learned Landscapes for Augmented Reality
06:54

Photorealistic Learned Landscapes for Augmented Reality

Published on: June 27, 2025

589
  • The objective function was modified to include a data-matching term on the network input.

    • Main Results:

    • The proposed GAN-based method demonstrated noticeable improvements in PET image reconstruction.
    • The method showed better lesion contrast recovery compared to kernel-based and U-net-based regularization.
    • Reduced background noise was observed with the proposed approach in both simulated and real patient data.

    Conclusions:

    • Generative adversarial networks offer a promising approach for regularized PET image reconstruction.
    • The modified objective function and GAN integration enhance image quality and diagnostic performance.
    • This method provides a superior trade-off between lesion contrast and background noise.