SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATION

Skin cancer, one of the most common and potentially fatal cancers, requires early and correct diagnosis to improve patient outcomes. While dermatologists possess extensive diagnostic expertise, recent studies have shown that Convolutional Neural Networks (CNNs) can often surpass human performance in...

Full description

Saved in:
Bibliographic Details
Main Authors: Maysaa R. Naeemah, Mohammed Y. Kamil
Format: Article
Language:English
Published: XLESCIENCE 2025-06-01
Series:International Journal of Advances in Signal and Image Sciences
Subjects:
Online Access:https://xlescience.org/index.php/IJASIS/article/view/276
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849709204945764352
author Maysaa R. Naeemah
Mohammed Y. Kamil
author_facet Maysaa R. Naeemah
Mohammed Y. Kamil
author_sort Maysaa R. Naeemah
collection DOAJ
description Skin cancer, one of the most common and potentially fatal cancers, requires early and correct diagnosis to improve patient outcomes. While dermatologists possess extensive diagnostic expertise, recent studies have shown that Convolutional Neural Networks (CNNs) can often surpass human performance in multiclass skin lesion classification, owing to their ability to extract subtle and complex features that may be overlooked during manual examination. In this study, SkinVGG-Net, an enhanced deep learning framework based on the VGG19 architecture is proposed. The SkinVGG-Net uses transfer learning and an advanced fine-tuning strategy, along with a comprehensive data preprocessing pipeline that includes image normalization, resizing, and extensive data augmentation to address the class imbalance challenges. The SkinVGG-Net model achieves 90.95% accuracy on HAM10000 dataset, with high performance across seven types of skin cancer. These results demonstrate the capability of CNN-based systems to provide consistent, accurate diagnostic support and highlight their promise as assistive tools in clinical settings with greater precision than traditional visual assessment methods.
format Article
id doaj-art-5c7540d50c0e4dfb9a5ba3961f818e83
institution DOAJ
issn 2457-0370
language English
publishDate 2025-06-01
publisher XLESCIENCE
record_format Article
series International Journal of Advances in Signal and Image Sciences
spelling doaj-art-5c7540d50c0e4dfb9a5ba3961f818e832025-08-20T03:15:23ZengXLESCIENCEInternational Journal of Advances in Signal and Image Sciences2457-03702025-06-0111116917910.29284/ijasis.11.1.2025.169-179304SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATIONMaysaa R. NaeemahMohammed Y. KamilSkin cancer, one of the most common and potentially fatal cancers, requires early and correct diagnosis to improve patient outcomes. While dermatologists possess extensive diagnostic expertise, recent studies have shown that Convolutional Neural Networks (CNNs) can often surpass human performance in multiclass skin lesion classification, owing to their ability to extract subtle and complex features that may be overlooked during manual examination. In this study, SkinVGG-Net, an enhanced deep learning framework based on the VGG19 architecture is proposed. The SkinVGG-Net uses transfer learning and an advanced fine-tuning strategy, along with a comprehensive data preprocessing pipeline that includes image normalization, resizing, and extensive data augmentation to address the class imbalance challenges. The SkinVGG-Net model achieves 90.95% accuracy on HAM10000 dataset, with high performance across seven types of skin cancer. These results demonstrate the capability of CNN-based systems to provide consistent, accurate diagnostic support and highlight their promise as assistive tools in clinical settings with greater precision than traditional visual assessment methods.https://xlescience.org/index.php/IJASIS/article/view/276deep learning, transfer learning, skin cancer, ai in healthcare, medical imaging.
spellingShingle Maysaa R. Naeemah
Mohammed Y. Kamil
SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATION
International Journal of Advances in Signal and Image Sciences
deep learning, transfer learning, skin cancer, ai in healthcare, medical imaging.
title SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATION
title_full SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATION
title_fullStr SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATION
title_full_unstemmed SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATION
title_short SKINVGG-NET: A MODIFIED AND FINE-TUNED VGG19-BASED DEEP LEARNING ARCHITECTURE FOR SKIN CANCER CLASSIFICATION
title_sort skinvgg net a modified and fine tuned vgg19 based deep learning architecture for skin cancer classification
topic deep learning, transfer learning, skin cancer, ai in healthcare, medical imaging.
url https://xlescience.org/index.php/IJASIS/article/view/276
work_keys_str_mv AT maysaarnaeemah skinvggnetamodifiedandfinetunedvgg19baseddeeplearningarchitectureforskincancerclassification
AT mohammedykamil skinvggnetamodifiedandfinetunedvgg19baseddeeplearningarchitectureforskincancerclassification