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: Nov 4, 2025

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.6K

Model-Based Deep Learning PET Image Reconstruction Using Forward-Backward Splitting Expectation-Maximization.

Abolfazl Mehranian1, Andrew J Reader1

  • 1School of Biomedical Engineering and Imaging Sciences, Department of Biomedical Engineering, King's College London, London SE1 7EH, U.K.

IEEE Transactions on Radiation and Plasma Medical Sciences
|May 31, 2021
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

Diffusion models for medical image reconstruction.

BJR artificial intelligence·2026
Same author

Synthesized Image Reconstruction for Post-Reconstruction Resolution Recovery.

IEEE transactions on radiation and plasma medical sciences·2024
Same author

AI for PET image reconstruction.

The British journal of radiology·2023
Same author

Deep Learning Signal Discrimination for Improved Sensitivity at High Specificity for CMOS Intraoperative Probes.

IEEE transactions on radiation and plasma medical sciences·2022
Same author

MR-guided motion-corrected PET image reconstruction for cardiac PET-MR.

Journal of nuclear medicine : official publication, Society of Nuclear Medicine·2021
Same author

Motion-corrected and high-resolution anatomically assisted (MOCHA) reconstruction of arterial spin labeling MRI.

Magnetic resonance in medicine·2020
Same journal

Impact of DOI Capability on Detector Performance: A Comparative Study of DOI and Non-DOI Detectors for High-Resolution and Sensitivity Organ-Specific PET Inserts.

IEEE transactions on radiation and plasma medical sciences·2026
Same journal

Evaluation of a PET Insert for Trimodal Imaging: A Step Toward PET/MRI-Guided Focused Ultrasound.

IEEE transactions on radiation and plasma medical sciences·2026
Same journal

CHEMONO: a Cherenkov-Only Monolithic Detector for PGI in Proton Range Verification.

IEEE transactions on radiation and plasma medical sciences·2026
Same journal

A Virtual Clinical Trial to Detect Changes in Tumor Uptake with PET using Lesion Embedding.

IEEE transactions on radiation and plasma medical sciences·2026
Same journal

Quantitative evaluation of spatially-variant deformations recovered by deep learning on clinical-like breast lesions.

IEEE transactions on radiation and plasma medical sciences·2026
Same journal

Impact of detector parameters and image resolution modeling on dedicated brain PET imaging.

IEEE transactions on radiation and plasma medical sciences·2026
See all related articles

Deep learning enhances Positron Emission Tomography (PET) image reconstruction. A novel algorithm, FBSEM, integrates deep learning with Expectation-Maximization for improved image quality, comparable to U-Net denoising.

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Computational Science

Background:

  • Maximum-a-posteriori (MAP) image reconstruction is crucial for Positron Emission Tomography (PET).
  • Integrating deep learning offers potential for improved PET image quality, especially in low-dose scenarios.
  • Existing methods like OSEM and Bowsher MAPEM have limitations in noise reduction and detail preservation.

Purpose of the Study:

  • To propose and evaluate a novel deep learning-based algorithm, FBSEM, for PET image reconstruction.
  • To compare the performance of FBSEM against traditional methods and U-Net denoising.
  • To assess the algorithm's effectiveness in both simulated and in-vivo low-dose PET data.

Main Methods:

  • Developed a forward-backward splitting EM (FBSEM) algorithm, unrolled into a recurrent neural network.
Keywords:
Deep learning (DL)MRIimage reconstructionpositron emission tomography (PET)

Related Experiment Videos

Last Updated: Nov 4, 2025

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.6K
  • Network parameters, including regularization strength, were learned during PET reconstruction.
  • Evaluated FBSEM-p (PET-only) and FBSEM-pm (PET-MR) on simulated and in-vivo brain imaging datasets.
  • Compared against OSEM, Bowsher MAPEM, and post-reconstruction U-Net denoising (Unet-p, Unet-pm).

    • Main Results:

    • FBSEM-p(m) and Unet-p(m) achieved comparable performance in simulations (14.4% and 13.4% NRMSE vs. 20.7% and 17.7% for OSEM/MAPEM).
    • For in-vivo data, FBSEM-p(m) showed the lowest error (3.9%) compared to Unet-p(m) (5.7%), MAPEM (5.9%), and OSEM (7.8%).
    • The U-Net denoising method demonstrated comparable performance to the FBSEM net.

    Conclusions:

    • The proposed FBSEM algorithm effectively integrates deep learning into PET image reconstruction.
    • FBSEM and U-Net denoising offer significant improvements over traditional methods for low-dose PET imaging.
    • Both deep learning approaches show promise for enhancing the quality of PET scans.