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: Jan 15, 2026

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.9K

Cycle-constrained adversarial denoising convolutional network for PET image denoising: Multi-dimensional validation

Yucun Hou1, Fenglin Zhan2, Jun Liu3

  • 1School of Physics and Astronomy, Key Laboratory of Beam Technology of Ministry of Education, Beijing Normal University, Beijing 100875, China.

Medical Image Analysis
|October 7, 2025
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

Long-cycling and High-voltage Solid State Lithium Metal Batteries Enabled by Fluorinated and Crosslinked Polyether Electrolytes.

Angewandte Chemie (International ed. in English)·2024
Same author

Succulent-Inspired Implicit Structural Change for Smart "ON/OFF" Switchable and Flexible EMI Shielding Coating.

ACS applied materials & interfaces·2024
Same author

Global research on RNA vaccines for COVID-19 from 2019 to 2023: a bibliometric analysis.

Frontiers in immunology·2024
Same author

Macular Neural and Microvascular Alterations in Type 2 Diabetes Without Retinopathy: A SS-OCT Study.

American journal of ophthalmology·2024
Same author

Tumor characteristics, brain functional activity, and connectivity of tinnitus in patients with vestibular schwannoma: a pilot study.

Quantitative imaging in medicine and surgery·2024
Same author

Dysregulation of ferroptosis-related genes in granulosa cells associates with impaired oocyte quality in polycystic ovary syndrome.

Frontiers in endocrinology·2024
This summary is machine-generated.

A new Cycle-DCN AI model reconstructs high-quality Positron Emission Tomography (PET) images from low-dose scans, reducing radiation risk for patients. This advanced denoising network preserves crucial details and diagnostic accuracy, validated across diverse datasets.

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Radiology

Background:

  • Positron Emission Tomography (PET) is vital for diagnosing diseases but involves radiation exposure.
  • Reducing radiation dose in PET scans often compromises image quality, limiting diagnostic potential.
  • Sensitive populations, like children, are particularly vulnerable to radiation risks associated with PET imaging.

Purpose of the Study:

  • To develop an advanced AI model for reconstructing high-quality PET images from low-dose scans.
  • To mitigate radiation risks in PET imaging without sacrificing diagnostic image fidelity.
  • To validate the proposed model's effectiveness across diverse patient demographics and scanner types.

Main Methods:

  • Proposed a Cycle-constrained Adversarial Denoising Convolutional Network (Cycle-DCN) integrating noise prediction, discriminators, and consistency networks.
Keywords:
Cycle-consistent adversarial networksImage denoisingPositron emission tomography (PET)

Related Experiment Videos

Last Updated: Jan 15, 2026

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.9K
  • Optimized the model using a combination of supervised, adversarial, cycle consistency, identity, and SSIM losses.
  • Trained and tested on a large dataset of 1224 brain PET scans and validated on 50 whole-body and 245 pediatric whole-body PET datasets from different scanners.

    • Main Results:

    • Cycle-DCN significantly improved Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Normalized Root Mean Square Error (NRMSE) by up to 56%, 35%, and 71%, respectively.
    • Restored images closely matched full-dose image quality in contrast-to-noise ratio (CNR) and edge preservation, maintaining tumor shape and detail.
    • Reader studies showed nuclear medicine physicians preferred Cycle-DCN reconstructed images, indicating strong clinical relevance and diagnostic utility.

    Conclusions:

    • The Cycle-DCN model effectively reconstructs diagnostic-quality PET images from low-dose scans, enhancing patient safety.
    • The AI approach preserves critical image details, tumor morphology, and contrast, addressing limitations of traditional low-dose PET reconstruction.
    • The model demonstrates generalizability across different PET scanners, patient populations, and anatomical regions, highlighting its clinical applicability.