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: Jun 28, 2026

Identifying, Diagnosing, and Grading Malignant Peripheral Nerve Sheath Tumors in Genetically Engineered Mouse Models
08:57

Identifying, Diagnosing, and Grading Malignant Peripheral Nerve Sheath Tumors in Genetically Engineered Mouse Models

Published on: May 17, 2024

Pathology-informed Generative Adversarial Network Augmentation Improves Classification of Peripheral Nerve Sheath

Giovanna Calabrese Dos Santos1, Hyago Vieira Lemes Barbosa Silva1,2, Anna Luíza Damaceno Araújo3

  • 1Institute of Science and Technology, Federal University of São Paulo (ICT- UNIFESP), São José dos Campos, São Paulo, Brazil.

Head and Neck Pathology
|June 26, 2026
PubMed

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

Salivary Methotrexate as a Predictor of Oral Mucositis in Hematological Cancer Patients: A Phase II Randomized, Double-Blind Clinical Trial.

Oral diseases·2026
Same author

Nomograms Predictive of Recurrence in Malignant Tumors Treated With Skull Base Surgery: An International Collaborative Study.

Head & neck·2026
Same author

Efficacy of Customized Toothbrush Handles in Individuals With Down Syndrome and Autism Spectrum Disorder: A Non-Randomized Clinical Trial.

Special care in dentistry : official publication of the American Association of Hospital Dentists, the Academy of Dentistry for the Handicapped, and the American Society for Geriatric Dentistry·2026
Same author

Non-Metastatic Squamous Cell Carcinoma of the Oropharynx: Primary Surgery or (Chemo)radiotherapy?

Head & neck·2026
Same author

MicroRNA expression signatures associated with metastatic progression in papillary thyroid carcinoma.

Archives of endocrinology and metabolism·2026
Same author

Beyond fragmented advances: rethinking recurrent head and neck cancer through integrated, cooperative research.

Journal of the National Cancer Institute·2026
Summary

Generative Adversarial Networks (GANs) enhance AI classification of head and neck peripheral nerve sheath tumors (PNSTs). Pathology-informed GAN augmentation improves accuracy, especially for rare perineurioma subtypes, aiding digital pathology diagnostics.

Area of Science:

  • Digital Pathology
  • Artificial Intelligence in Oncology
  • Computational Pathology

Background:

  • Head and neck peripheral nerve sheath tumors (PNSTs) exhibit significant histopathological overlap.
  • Convolutional Neural Networks (CNNs) show promise for soft tissue tumor classification but struggle with rare subtypes like perineurioma due to limited intra-class variability.

Purpose of the Study:

  • To evaluate the efficacy of Generative Adversarial Network (GAN)-based synthetic data augmentation in improving CNN classification accuracy for PNSTs.
  • To assess the impact of morphology-driven augmentation strategies on the classification of rare and morphologically heterogeneous tumor subtypes.

Main Methods:

  • A retrospective study involved 30 patients with PNSTs; whole-slide images were digitized and split for training and testing.
Keywords:
Convolutional neural networksGenerative adversarial networksImage augmentationPerineuriomaPeripheral nerve sheath tumorsSynthetic patch generationWhole-slide images

More Related Videos

An Orthotopic Sciatic Nerve Xenograft for Neurofibromatosis Type 1 Neurofibromas
03:53

An Orthotopic Sciatic Nerve Xenograft for Neurofibromatosis Type 1 Neurofibromas

Published on: October 10, 2025

Related Experiment Videos

Last Updated: Jun 28, 2026

Identifying, Diagnosing, and Grading Malignant Peripheral Nerve Sheath Tumors in Genetically Engineered Mouse Models
08:57

Identifying, Diagnosing, and Grading Malignant Peripheral Nerve Sheath Tumors in Genetically Engineered Mouse Models

Published on: May 17, 2024

An Orthotopic Sciatic Nerve Xenograft for Neurofibromatosis Type 1 Neurofibromas
03:53

An Orthotopic Sciatic Nerve Xenograft for Neurofibromatosis Type 1 Neurofibromas

Published on: October 10, 2025

  • Synthetic perineurioma patches were generated using a modified Pix2Pix GAN with bottleneck architecture and self-attention.
  • Two augmentation strategies were tested: intra-phenotypic expansion (within-subtype) and inter-phenotypic interpolation (between-subtypes). EfficientNetV2-B0 was trained on original data, original+Experiment A, and original+Experiment B datasets.

    • Main Results:

    • GAN-based augmentation significantly improved global classification accuracy (0.733 baseline vs. augmented models).
    • Intra-phenotypic expansion yielded the highest balanced accuracy (0.750) and macro-F1 score (0.740), improving perineurioma recall from 0.34 to 0.51.
    • Inter-phenotypic interpolation achieved the highest overall accuracy and improved multiclass agreement, with specificity remaining high (≥0.999).

    Conclusions:

    • Pathology-informed GAN augmentation effectively enhances CNN performance for PNST classification, particularly for challenging perineurioma cases.
    • Intra-phenotypic expansion boosts sensitivity for rare classes, while inter-phenotypic interpolation enhances global robustness and multiclass agreement.
    • Morphology-driven synthetic data enrichment is a valuable strategy for improving AI in digital pathology, especially for underrepresented tumor entities.