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 Videos

Multi-modality brain tumor segmentation using dual-attention generative adversarial network.

Daisuke Kawahara1, Takahiro Nishimura1, Masato Tsuneda2

  • 1Department of Radiation Oncology, Graduate School of Biomedical Health Sciences, Hiroshima University, Hiroshima, Japan.

Reports of Practical Oncology and Radiotherapy : Journal of Greatpoland Cancer Center in Poznan and Polish Society of Radiation Oncology
|July 14, 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

Development of an End-to-end Phantom and Procedure for Computed Tomography and Cone-beam Computed Tomography-guided Online Adaptive Radiotherapy.

Journal of medical physics·2026
Same author

In vivo dosimetry of cardiac implantable electronic devices with a radiophotoluminescent glass dosimeter in patients undergoing radiotherapy.

Radiological physics and technology·2026
Same author

Development of a virtual phantom for HDR cervical cancer brachytherapy: A multi-institutional tool for training, quality assurance, and variability reduction.

Journal of applied clinical medical physics·2026
Same author

Evaluation of segmentation accuracy and the improvement of time effectiveness using deep learning-based segmentation in <sup>177</sup>Lu-DOTATATE dosimetry.

EJNMMI physics·2026
Same author

Credentialing of small-field dosimetry for the lung SBRT in Japan Clinical Oncology Group clinical trial.

Journal of applied clinical medical physics·2026
Same author

Survival prediction in non-small cell lung cancer using layer-wise radiomics and stacked radscore integration: a multi-institutional study.

Physica medica : PM : an international journal devoted to the applications of physics to medicine and biology : official journal of the Italian Association of Biomedical Physics (AIFB)·2026

A new dual-attention generative adversarial network (DAtGAN) model significantly improves brain tumor segmentation accuracy on MRI scans. This advanced deep learning approach enhances glioma segmentation, aiding clinical diagnostics and treatment planning.

Area of Science:

  • Medical imaging analysis
  • Artificial intelligence in healthcare
  • Neuro-oncology

Background:

  • Accurate segmentation of malignant brain tumors is crucial for diagnostics and radiation therapy.
  • Multi-contrast MRI is used to delineate tumor subregions like necrotic, enhancing, and non-enhancing areas.
  • Existing auto-segmentation models require improvement for clinical utility.

Purpose of the Study:

  • To enhance an auto-segmentation model for glioma using a generative adversarial network (GAN) with dual-attention (DAtGAN).
  • To evaluate the DAtGAN model's performance on multi-contrast MRI scans for brain tumor segmentation.
  • To compare the DAtGAN model against conventional GANs for segmentation accuracy.

Main Methods:

  • Utilized the Brain Tumor Segmentation (BraTS) 2017 challenge dataset.
Keywords:
attentioncritical organssegmentation

Related Experiment Videos

  • Developed a DAtGAN model incorporating an attention module to leverage global information for both generator and discriminator.
  • Compared segmentation performance metrics (Dice Similarity Coefficient, Intersection Over the Union, Hausdorff Distance) against a conventional GAN.

    • Main Results:

    • DAtGAN achieved higher average Dice Similarity Coefficients (DSC) compared to the conventional GAN: 0.88 (enhancing tumor), 0.92 (core tumor), and 0.91 (whole tumor).
    • The DAtGAN model demonstrated superior performance with higher Intersection Over the Union and lower maximum Hausdorff distance.
    • The DAtGAN model showed a notable improvement over the GAN model across all evaluated tumor regions.

    Conclusions:

    • The DAtGAN model effectively improves auto-segmentation of glioma patients using MRI.
    • The proposed DAtGAN shows potential for superior segmentation performance compared to other deep learning and atlas-based methods.
    • DAtGAN offers a valuable tool for clinical settings, enhancing segmentation accuracy and efficiency for glioma patients, thereby improving treatment planning and outcomes.