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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

Epidemiological and clinical features of hypertrophic scar and keloid in Chinese college students: A university-based cross-sectional survey.

Heliyon·2023
Same author

Five new terpenoids from Viburnum odoratissimum var. sessiliflorum.

Chinese journal of natural medicines·2023
Same author

Characterization of a Riboflavin-Producing Mutant of <i>Bacillus subtilis</i> Isolated by Droplet-Based Microfluidics Screening.

Microorganisms·2023
Same author

Neuroprotection of SRT2104 in Murine Ischemia/Reperfusion Injury Through the Enhancement of Sirt1-Mediated Deacetylation.

Investigative ophthalmology & visual science·2023
Same author

SIRT6 ameliorates LPS-induced apoptosis and tight junction injury in ARDS through the ERK1/2 pathway and autophagy.

International journal of medical sciences·2023
Same author

Deep learning-based diagnosis of osteoblastic bone metastases and bone islands in computed tomograph images: a multicenter diagnostic study.

European radiology·2023

Related Experiment Video

Updated: Jun 26, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.7K

Unsupervised low-dose CT denoising using bidirectional contrastive network.

Yuanke Zhang1, Rui Zhang2, Rujuan Cao2

  • 1School of Computer Science, Qufu Normal University, Rizhao 276826, China; Shandong Provincial Key Laboratory of Data Security and Intelligent Computing, Qufu Normal University, Rizhao 276826, China.

Computer Methods and Programs in Biomedicine
|May 9, 2024
PubMed
Summary
This summary is machine-generated.

This study introduces an unsupervised method for denoising low-dose computed tomography (LDCT) images, improving quality without needing paired data. The Bidirectional Contrastive Unsupervised Denoising (BCUD) model enhances diagnostic accuracy and clinical applicability.

Keywords:
Bidirectional network structureContrastive learningImage denoisingLow-dose CTUnsupervised learning

More Related Videos

Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
14:08

Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images

Published on: April 13, 2013

42.6K
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

522

Related Experiment Videos

Last Updated: Jun 26, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.7K
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
14:08

Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images

Published on: April 13, 2013

42.6K
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

522

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Image Processing

Background:

  • Low-dose computed tomography (LDCT) reduces radiation but introduces noise and artifacts, compromising image quality and diagnostic accuracy.
  • Supervised learning methods for LDCT denoising require large, paired datasets, posing significant acquisition challenges.

Purpose of the Study:

  • To develop a robust unsupervised LDCT denoising method that overcomes the reliance on paired LDCT and normal-dose CT (NDCT) samples.
  • To improve the accessibility and practical application of LDCT denoising techniques.

Main Methods:

  • Propose a novel unsupervised network model, Bidirectional Contrastive Unsupervised Denoising (BCUD).
  • Utilize a bidirectional network structure with contrastive learning to map correspondences between noisy LDCT and clean NDCT domains.
  • Employ dual encoders, discriminators, and unique projection heads for domain-specific data generation and adaptive feature representation.
  • Align corresponding features across domains in learned embedding spaces for noise reduction and detail preservation.

Main Results:

  • BCUD demonstrated superior performance on public and clinical datasets.
  • Achieved high quantitative metrics: PSNR (31.387 dB), SSIM (0.886), IFC (2.305), VIF (0.373).
  • Radiologists' subjective evaluation yielded a mean score of 4.23, indicating strong clinical applicability.

Conclusions:

  • An innovative unsupervised LDCT denoising method using a bidirectional contrastive network was presented.
  • The BCUD method significantly improves clinical applicability by eliminating the need for matched image pairs.
  • This approach sets a new benchmark in unsupervised LDCT denoising, excelling in noise reduction and fine detail preservation.