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 Video

Updated: Dec 24, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Transfer learning using a multi-scale and multi-network ensemble for skin lesion classification.

Amirreza Mahbod¹, Gerald Schaefer², Chunliang Wang³

¹Institute for Pathophysiology and Allergy Research, Medical University of Vienna, Vienna, Austria; Research and Development Department of TissueGnostics GmbH, Vienna, Austria.

Computer Methods and Programs in Biomedicine

|April 9, 2020

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

Editorial: Artificial intelligence advancing lung cancer screening and treatment.

Frontiers in oncology·2026

Same author

Two years of COVID-19: persistently reduced well-being and increases in global psychopathology during the pandemic in a representative Austrian population-sample within the COH-FIT study.

Frontiers in psychiatry·2026

Same author

Editorial: Innovative organoid co-culture systems for enhanced precision medicine in cancer and beyond.

Frontiers in medicine·2026

Same author

A fusion-based multiomics classification approach for enhanced gene discovery in non-small cell lung cancer.

Bioinformatics advances·2026

Same author

Negative symptoms in schizophrenia: Methodological challenges and emerging solutions for outcome measurement.

Neuroscience applied·2026

Same author

Expert Opinions on Postoperative Complications in Breast Cancer Surgery After Neoadjuvant Chemotherapy: A Descriptive Study Through Structured Interviews With Surgeons in Austria.

The breast journal·2026

Same journal

Facial iPPG heatmap patterns based on period-aware autoencoder show association with carotid atherosclerosis towards non-contact hemodynamic assessment.

Computer methods and programs in biomedicine·2026

Same journal

Explainable machine learning models predict liver fibrosis risk and outcome in the general population: Development and multi-cohort external validation.

Computer methods and programs in biomedicine·2026

Same journal

Evaluation of surrogate endpoints for survival outcomes using the surrogate package in R.

Computer methods and programs in biomedicine·2026

Same journal

Relative spectral and frication-based descriptors as numerical indicators of place of articulation shifts in fricatives produced by Polish children.

Computer methods and programs in biomedicine·2026

Same journal

Leaflet resection improves valve expansion and hemodynamic performance in redo TAVI with balloon- and self-expanding transcatheter heart valve configurations.

Computer methods and programs in biomedicine·2026

Same journal

Spectral super-resolution for Parkinson's voice via representation-level methods under mixed-reality acquisition.

Computer methods and programs in biomedicine·2026

See all related articles

This summary is machine-generated.

Image cropping and multi-CNN fusion improve skin cancer classification. This study demonstrates that cropping dermoscopic images and using a multi-scale multi-CNN approach enhances diagnostic accuracy for skin lesions.

Area of Science:

Dermatology
Computer Science
Artificial Intelligence

Background:

Skin cancer is a prevalent cancer type in fair-skinned populations.
Computer-aided diagnosis of skin lesions using dermoscopic images is of significant interest.
Transfer learning with pre-trained convolutional neural networks (CNNs) shows promise for skin lesion classification.

Purpose of the Study:

To investigate the impact of image size on skin lesion classification using transfer learning.
To compare image resizing versus cropping strategies for dermoscopic images.
To evaluate a multi-scale multi-CNN (MSM-CNN) fusion approach for enhanced classification performance.

Main Methods:

Dermoscopic images from the ISIC dataset were resized or cropped to various sizes (224x224 to 450x450).

Keywords:

Deep learning Dermoscopy Image cropping Image resolution Medical image analysis Skin cancer Transfer learning

Related Experiment Videos

Last Updated: Dec 24, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Three established CNNs (EfficientNetB0, EfficientNetB1, SeReNeXt-50) were fine-tuned and evaluated.

A novel MSM-CNN fusion approach using an ensemble strategy with cropped images at multiple scales was proposed and tested.

Main Results:

Image cropping outperformed image resizing, yielding superior classification performance across all scales.
The proposed MSM-CNN fusion approach significantly boosted classification performance compared to single networks or scales.
The MSM-CNN algorithm achieved a balanced multi-class accuracy of 86.2% on the ISIC 2018 test set, ranking second on the live leaderboard.

Conclusions:

Image size critically affects skin lesion classification performance in CNN transfer learning.
Cropping dermoscopic images is a more effective strategy than resizing for improving classification accuracy.
A multi-scale ensemble of fine-tuned CNNs using cropped images offers the best performance for skin lesion diagnosis.