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: Jul 5, 2026

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

RGCNN-nnUNet: Recurrent group equivariant nnU-Net for robust brain tissue segmentation on stroke NCCT.

Xiang Li1, Fengyuan Wang2, Saurabh Bagchi1

  • 1Purdue University, West Lafayette, 47906, United States.

Computerized Medical Imaging and Graphics : the Official Journal of the Computerized Medical Imaging Society
|July 3, 2026
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

Neural checkpoint therapy in lung cancer.

Trends in cancer·2026
Same author

Quantitative second harmonic generation microscopy for characterizing collagen remodeling in papillary thyroid carcinoma.

Journal of biomedical optics·2026
Same author

Stimuli-responsive microneedles: Engineered solutions for multifaceted diabetic wound healing.

Journal of applied biomaterials & functional materials·2026
Same author

iS2C2: a cointelligent platform for mechanistic discovery of disease cellular crosstalk.

Signal transduction and targeted therapy·2026
Same author

Graphene quantum dot-coated mesoporous silica loaded with glucosamine sulfate as a potent antigen delivery system for immunoactivation.

Communications biology·2026
Same author

Obesity-driven phosphatidylethanolamine dysregulation impairs neuroimmune crosstalk and accelerates Alzheimer's pathogenesis.

Molecular neurodegeneration·2026

A new AI model, RGCNN-nnUNet, enhances brain tissue segmentation for early stroke detection using non-contrast CT (NCCT). This improves diagnostic sensitivity by 29%, aiding in the identification of subtle infarct signals.

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Neurology

Background:

  • Non-contrast CT (NCCT) is vital for emergency stroke diagnosis but struggles with subtle early ischemic changes due to noise and low contrast.
  • Accurate brain tissue segmentation is essential for identifying these subtle signals.

Purpose of the Study:

  • To develop and evaluate RGCNN-nnUNet, a novel deep learning architecture for high-fidelity brain tissue segmentation on NCCT.
  • To assess the clinical impact of RGCNN-nnUNet in improving stroke diagnosis sensitivity.

Main Methods:

  • Introduced a Recurrent Group Convolutional Cell leveraging group equivariance for consistent feature extraction and stable refinement.
  • Evaluated RGCNN-nnUNet on multi-center datasets for brain tissue segmentation.
  • Conducted a blinded reader study comparing AI-denoised NCCT with RGCNN-nnUNet segmentation maps against standard radiology reports.
Keywords:
Brain tissue segmentationDeep neural networkNon-contrast CTRecurrent group equivariant CNN

Related Experiment Videos

Last Updated: Jul 5, 2026

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

Main Results:

  • RGCNN-nnUNet achieved state-of-the-art segmentation performance with Dice coefficients of 0.885 (white matter) and 0.86 (deep gray matter).
  • Demonstrated significant reductions in Hausdorff Distance (HD95) by 14.3% (white matter) and 10.6% (deep gray matter).
  • Integration with AI-denoised NCCT increased diagnostic sensitivity by 29% compared to standard reports.

Conclusions:

  • RGCNN-nnUNet provides precise anatomical boundaries and stable tissue-weighting for improved brain tissue segmentation.
  • The framework significantly enhances diagnostic sensitivity for subtle infarct signals in early stroke detection.
  • This approach holds potential for improving emergency stroke diagnosis workflows.