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: Oct 9, 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

3.0K

Single Model Deep Learning on Imbalanced Small Datasets for Skin Lesion Classification.

Peng Yao, Shuwei Shen, Mengjuan Xu

    IEEE Transactions on Medical Imaging
    |December 20, 2021
    PubMed
    Summary
    This summary is machine-generated.

    Related Concept Videos

    Skin Cancer01:30

    Skin Cancer

    4.9K
    Skin cancer is a type of cancer that occurs when there is an abnormal growth of skin cells, usually triggered by damage to the DNA within the skin cells. It is primarily caused by exposure to ultraviolet (UV) radiation from the sun or artificial sources like tanning beds. Skin cancer is the most common type of cancer worldwide, and its incidence continues to rise.
    Basal Cell Carcinoma (BCC): BCC is the most common type of skin cancer, accounting for about 80% of cases. It typically develops in...
    4.9K

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Multiomic characterization, early detection, and therapeutic targeting of myeloid sarcoma.

    bioRxiv : the preprint server for biology·2026
    Same author

    Long-term morbidity and quality of life in survivors of Stevens-Johnson syndrome/toxic epidermal necrolysis spectrum.

    The British journal of dermatology·2025
    Same author

    HLA-B Alleles With Shared Peptide Binding Specificities Define Global Risk of Co-trimoxazole-Induced Severe Cutaneous Adverse Drug Reactions.

    The journal of allergy and clinical immunology. In practice·2025
    Same author

    Automating the Diagnosis of Ulcerative Skin Disorders: Google's Teachable Machine Can Identify Validated Cases of Pyoderma Gangrenosum.

    International journal of dermatology·2025
    Same author

    HLA-B Alleles with Shared Peptide Binding Specificities Define Global Risk of Cotrimoxazole-induced SCAR.

    medRxiv : the preprint server for health sciences·2025
    Same author

    Electronic patient message burdens: An analysis of factors associated with electronic patient message quantity and turnaround time in dermatology.

    Journal of the American Academy of Dermatology·2025

    This study introduces a single deep convolutional neural network (DCNN) model for skin lesion classification on small, imbalanced datasets. The method achieves high accuracy, comparable to ensemble models, using novel augmentation and loss functions.

    Area of Science:

    • Dermatology
    • Artificial Intelligence
    • Medical Imaging

    Background:

    • Deep convolutional neural network (DCNN) models show promise in skin disease diagnosis, often matching or exceeding dermatologist accuracy.
    • However, limited dataset size and imbalance in public skin lesion datasets hinder widespread DCNN implementation.
    • Existing methods struggle with small, imbalanced datasets, limiting practical application in skin lesion detection.

    Purpose of the Study:

    • To propose a novel single-model strategy for effective skin lesion classification on small and imbalanced datasets.
    • To address challenges of overfitting, underrepresentation, and uneven sample sizes in dermoscopic image datasets.
    • To develop a computationally efficient DCNN approach suitable for mobile-based automated skin lesion screening.

    Main Methods:

    Related Experiment Videos

    Last Updated: Oct 9, 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

    3.0K
    • Trained various DCNNs on small, imbalanced datasets, observing that moderately complex models performed better than larger ones.
    • Incorporated DropOut and DropBlock regularization to mitigate overfitting.
    • Proposed a Modified RandAugment strategy to address sample underrepresentation.
    • Introduced a Multi-Weighted New Loss (MWNL) function and a cumulative learning strategy (CLS) to handle sample imbalance and classification difficulty.

    Main Results:

    • The proposed single DCNN model, utilizing Modified RandAugment, MWNL, and CLS, achieved classification accuracy comparable or superior to multiple ensemble models.
    • Demonstrated that moderately complex DCNNs, with appropriate regularization and augmentation, can outperform larger models on limited data.
    • The combined strategy effectively overcame challenges posed by small and imbalanced datasets.

    Conclusions:

    • The developed single DCNN model offers a high-performance, cost-effective solution for skin lesion classification, even with limited data.
    • This approach is suitable for deployment on mobile devices for automated skin lesion screening, particularly in resource-limited settings.
    • The study highlights the potential of tailored DCNN strategies to overcome data limitations in medical image analysis.