Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Melanoma detection using adversarial training and deep transfer learning.

Hasib Zunair¹, A Ben Hamza¹

¹Concordia University Montreal QC Canada.

Physics in Medicine and Biology

|April 7, 2020

Summary

This summary is machine-generated.

Related Concept Videos

Skin Cancer

Skin Cancer

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...

You might also read

Related Articles

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

Sort by

Same author

CosSIF: Cosine similarity-based image filtering to overcome low inter-class variation in synthetic medical image datasets.

Computers in biology and medicine·2024

Same author

Regular Splitting Graph Network for 3D Human Pose Estimation.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2023

Same author

Quantifying imbalanced classification methods for leukemia detection.

Computers in biology and medicine·2022

Same author

Knowledge distillation approach towards melanoma detection.

Computers in biology and medicine·2022

Same author

Automatic segmentation of blood cells from microscopic slides: A comparative analysis.

Tissue & cell·2021

Same author

Sharp U-Net: Depthwise convolutional network for biomedical image segmentation.

Computers in biology and medicine·2021

Same journal

Deep learning-based dose prediction to enhance planning efficiency in cervical brachytherapy with hybrid applicators.

Physics in medicine and biology·2026

Same journal

Novel TMS coils designed using an inverse boundary element method.

Physics in medicine and biology·2026

Same journal

PAC-Net: patch adaptive cut-off network with differentiable module-wise K-learning for robust and efficient medical image segmentation.

Physics in medicine and biology·2026

Same journal

Four-dimensional on-beam computed tomography reconstruction using projection-difference images.

Physics in medicine and biology·2026

Same journal

Higher-order synergy-based ranking in transcriptomic communities via latent factors and O-information.

Physics in medicine and biology·2026

Same journal

Calculating biological dose distributions in hadrontherapy using GATE: the BioDose actor.

Physics in medicine and biology·2026

See all related articles

This study introduces a novel two-stage framework to improve melanoma detection by synthesizing under-represented skin lesion images. The method enhances classification accuracy, potentially matching expert dermatologist performance.

Area of Science:

Medical Imaging
Artificial Intelligence in Dermatology
Computational Pathology

Background:

Skin lesion datasets suffer from class imbalance, with few abnormal samples compared to normal ones.
Low inter-class variability in skin lesion images complicates accurate classification.
Accurate melanoma detection is crucial for early diagnosis and improved patient outcomes.

Purpose of the Study:

To propose a two-stage framework for automatic skin lesion image classification.
To address class imbalance and low inter-class variability in skin lesion datasets.
To enhance the accuracy of melanoma detection through advanced machine learning techniques.

Main Methods:

A two-stage framework combining adversarial training and transfer learning for melanoma detection.

Related Experiment Videos

Stage 1: Conditional image synthesis using unpaired image-to-image translation to generate under-represented class samples.

Stage 2: Training a deep convolutional neural network classifier with original and synthesized data, using focal loss to handle hard examples.

Main Results:

The proposed approach significantly outperformed standard baseline methods on a dermatology image benchmark.
Achieved substantial performance improvements in skin lesion classification accuracy.
Feature visualization indicated context-based lesion assessment capabilities comparable to expert dermatologists.

Conclusions:

The novel two-stage framework effectively addresses class imbalance and low inter-class variability in skin lesion datasets.
The method demonstrates superior performance in automatic skin lesion classification and melanoma detection.
The approach shows potential for aiding dermatologists in lesion assessment, reaching expert levels of accuracy.