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

VIVIE: Virtually Integrated Ventricular Intervention Environment and its effectiveness as a teaching and learning tool.

International journal of computer assisted radiology and surgery·2026
Same author

Deep learning for synovial volume segmentation of the first carpometacarpal joint in osteoarthritis patients.

Osteoarthritis imaging·2026
Same author

Hippocampal volume for prediction of working memory performance in patients with infantile hydrocephalus.

Journal of neurosurgery. Pediatrics·2026
Same author

Domain-agnostic Unsupervised Domain Adaptation Segmentation from 3D Carotid Artery Ultrasound Image.

IEEE journal of biomedical and health informatics·2026
Same author

Automated extraction of the plane of minimal hiatal dimensions and mid-sagittal plane from 3D transperineal ultrasound.

Medical physics·2026
Same author

Subcortical morphological alterations and volume loss in infantile hydrocephalus: A surface- and volume-based analysis.

Neuroimage. Reports·2026

Related Experiment Video

Updated: Jun 24, 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

Semi-supervised learning framework with shape encoding for neonatal ventricular segmentation from 3D ultrasound.

Zachary Szentimrey1, Abdullah Al-Hayali1, Sandrine de Ribaupierre2

  • 1School of Engineering, University of Guelph, Guelph, Ontario, Canada.

Medical Physics
|June 10, 2024
PubMed
Summary

This study introduces a novel semi-supervised learning (SSL) method for segmenting neonatal brain ventricles in 3D ultrasound images, significantly improving accuracy and efficiency. The developed SSL approach effectively utilizes unlabeled data, outperforming existing methods in accuracy and computational demands.

Keywords:
3D ultrasoundbrain segmentationneonatal brainsemi‐supervised learning

More Related Videos

Modeling Neonatal Intraventricular Hemorrhage Through Intraventricular Injection of Hemoglobin
07:57

Modeling Neonatal Intraventricular Hemorrhage Through Intraventricular Injection of Hemoglobin

Published on: August 25, 2022

2.9K
Three-Dimensional Shape Modeling and Analysis of Brain Structures
05:33

Three-Dimensional Shape Modeling and Analysis of Brain Structures

Published on: November 14, 2019

7.1K

Related Experiment Videos

Last Updated: Jun 24, 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
Modeling Neonatal Intraventricular Hemorrhage Through Intraventricular Injection of Hemoglobin
07:57

Modeling Neonatal Intraventricular Hemorrhage Through Intraventricular Injection of Hemoglobin

Published on: August 25, 2022

2.9K
Three-Dimensional Shape Modeling and Analysis of Brain Structures
05:33

Three-Dimensional Shape Modeling and Analysis of Brain Structures

Published on: November 14, 2019

7.1K

Area of Science:

  • Medical imaging
  • Artificial intelligence in medicine
  • Neonatal neuroimaging

Background:

  • Three-dimensional (3D) ultrasound (US) is promising for monitoring neonatal intraventricular hemorrhaging.
  • Accurate segmentation of lateral ventricles in 3D US is challenging due to poor boundaries and low signal-to-noise ratio.
  • Fully supervised methods require extensive physician annotations, leading to high costs and potential overfitting.

Purpose of the Study:

  • To develop a fast, lightweight, and accurate semi-supervised learning (SSL) method for 3D US image segmentation.
  • To leverage unlabeled data to enhance segmentation performance for neonatal cerebral lateral ventricles.

Main Methods:

  • Proposed an SSL framework combining an autoencoder for shape-encoding and a 3D U-Net segmentation model.
  • Utilized pseudo-labels from the autoencoder to enforce shape constraints and an adversarial discriminator for data distribution analysis.
  • Experimented with 887 3D US images (87 labeled, 800 unlabeled) and compared performance using Dice Similarity Coefficient (DSC), Mean Absolute Surface Distance (MAD), and Absolute Volumetric Difference (VD).

Main Results:

  • The shape-encoding SSL method achieved mean DSC improvements of 6.5%, 7.7%, and 4.1% over the baseline 3D U-Net across different data splits.
  • The method demonstrated computational efficiency, with only a 1GB increase in RAM compared to the baseline and less RAM/parameters than a 3D U-Net ensemble.
  • Statistical significance was confirmed using the Wilcoxon signed-rank test with Bonferroni correction (p < 0.01).

Conclusions:

  • This work presents one of the first SSL methods specifically for 3D US organ segmentation, incorporating unlabeled data for neonatal cerebral lateral ventricles.
  • The proposed SSL method achieved superior DSC and lower VD compared to state-of-the-art SSL and fully supervised methods.
  • The approach is computationally efficient, offering a practical solution for medical image segmentation challenges.