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

3-To-1 Pipeline: Restructuring Transfer Learning Pipelines for Medical Imaging Classification via Optimized GAN

Ross Zhi Jian Choong, Seth Austin Harding, Bo-Yen Tang

    Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference
    |October 6, 2020
    PubMed
    Summary
    This summary is machine-generated.

    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

    Progressive Bounded Error Piecewise Linear Approximation with Resolution Reduction for Time Series Data Compression.

    Sensors (Basel, Switzerland)·2025
    Same author

    A Deep Learning-Based Chinese Semantic Parser for the Almond Virtual Assistant.

    Sensors (Basel, Switzerland)·2022
    Same author

    Real-Time Identification of Irrigation Water Pollution Sources and Pathways with a Wireless Sensor Network and Blockchain Framework.

    Sensors (Basel, Switzerland)·2020
    Same author

    Governance on the Drug Supply Chain via Gcoin Blockchain.

    International journal of environmental research and public health·2018

    This study introduces a novel training pipeline using generative models to synthesize medical images, overcoming limitations of traditional transfer learning for deep learning in medical imaging. This approach improves diagnostic accuracy by bridging the modality gap.

    Area of Science:

    • Medical Imaging
    • Deep Learning
    • Computer Vision

    Background:

    • Deep learning models require extensive training data, which is often scarce in biomedical imaging.
    • Current methods pre-train models on non-medical datasets like ImageNet, facing a modality gap (RGB vs. grayscale) that limits effectiveness in medical applications.
    • Generative models, specifically Generative Adversarial Networks (GANs), can synthesize new medical images.

    Purpose of the Study:

    • To address the challenge of limited training data in medical imaging.
    • To overcome the modality gap inherent in transfer learning from non-medical datasets.
    • To propose and evaluate a novel training pipeline utilizing GAN-generated medical images.

    Main Methods:

    • Development of a training pipeline incorporating Generative Adversarial Networks (GANs) for synthetic medical image generation.

    Related Experiment Videos

  • Comparison of the proposed pipeline against conventional GAN-synthetic methods.
  • Evaluation against established transfer learning techniques using non-medical datasets.

    • Main Results:

    • The proposed training pipeline demonstrated superior performance compared to conventional GAN-synthetic methods.
    • The pipeline outperformed traditional transfer learning approaches in medical imaging tasks.
    • Synthesized medical images effectively bridged the modality gap, enhancing model training.

    Conclusions:

    • The proposed training pipeline offers a viable solution to the data scarcity problem in medical deep learning.
    • Utilizing GAN-generated images is an effective strategy to overcome the modality gap in transfer learning.
    • This approach holds significant potential for improving the application of deep learning in biomedical imaging analysis.