Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer Classification

<b>Background</b>: Breast cancer remains one of the leading causes of mortality among women worldwide, highlighting the critical need for accurate and efficient diagnostic methods. <b>Methods</b>: Traditional deep learning models often struggle with feature redundancy, subopt...

Full description

Saved in:
Bibliographic Details
Main Authors: Soaad Ahmed, Naira Elazab, Mostafa M. El-Gayar, Mohammed Elmogy, Yasser M. Fouda
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Diagnostics
Subjects:
Online Access:https://www.mdpi.com/2075-4418/15/11/1361
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850160527944187904
author Soaad Ahmed
Naira Elazab
Mostafa M. El-Gayar
Mohammed Elmogy
Yasser M. Fouda
author_facet Soaad Ahmed
Naira Elazab
Mostafa M. El-Gayar
Mohammed Elmogy
Yasser M. Fouda
author_sort Soaad Ahmed
collection DOAJ
description <b>Background</b>: Breast cancer remains one of the leading causes of mortality among women worldwide, highlighting the critical need for accurate and efficient diagnostic methods. <b>Methods</b>: Traditional deep learning models often struggle with feature redundancy, suboptimal feature fusion, and inefficient selection of discriminative features, leading to limitations in classification performance. To address these challenges, we propose a new deep learning framework that leverages MAX-ViT for multi-scale feature extraction, ensuring robust and hierarchical representation learning. A gated attention fusion module (GAFM) is introduced to dynamically integrate the extracted features, enhancing the discriminative power of the fused representation. Additionally, we employ Harris Hawks optimization (HHO) for feature selection, reducing redundancy and improving classification efficiency. Finally, XGBoost is utilized for classification, taking advantage of its strong generalization capabilities. <b>Results</b>: We evaluate our model on the King Abdulaziz University Mammogram Dataset, categorized based on BI-RADS classifications. Experimental results demonstrate the effectiveness of our approach, achieving 98.2% for accuracy, 98.0% for precision, 98.1% for recall, 98.0% for F1-score, 98.9% for the area under the curve (AUC), and 95% for the Matthews correlation coefficient (MCC), outperforming existing state-of-the-art models. <b>Conclusions</b>: These results validate the robustness of our fusion-based framework in improving breast cancer diagnosis and classification.
format Article
id doaj-art-b6476d1322bd4cf0b38437ed1cb2a1bc
institution OA Journals
issn 2075-4418
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Diagnostics
spelling doaj-art-b6476d1322bd4cf0b38437ed1cb2a1bc2025-08-20T02:23:08ZengMDPI AGDiagnostics2075-44182025-05-011511136110.3390/diagnostics15111361Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer ClassificationSoaad Ahmed0Naira Elazab1Mostafa M. El-Gayar2Mohammed Elmogy3Yasser M. Fouda4Computer Science Division, Mathematics Department, Faculty of Science, Mansoura University, Mansoura 35516, EgyptInformation Technology Department, Faculty of Computers and Information, Mansoura University, Mansoura 35516, EgyptInformation Technology Department, Faculty of Computers and Information, Mansoura University, Mansoura 35516, EgyptInformation Technology Department, Faculty of Computers and Information, Mansoura University, Mansoura 35516, EgyptComputer Science Division, Mathematics Department, Faculty of Science, Mansoura University, Mansoura 35516, Egypt<b>Background</b>: Breast cancer remains one of the leading causes of mortality among women worldwide, highlighting the critical need for accurate and efficient diagnostic methods. <b>Methods</b>: Traditional deep learning models often struggle with feature redundancy, suboptimal feature fusion, and inefficient selection of discriminative features, leading to limitations in classification performance. To address these challenges, we propose a new deep learning framework that leverages MAX-ViT for multi-scale feature extraction, ensuring robust and hierarchical representation learning. A gated attention fusion module (GAFM) is introduced to dynamically integrate the extracted features, enhancing the discriminative power of the fused representation. Additionally, we employ Harris Hawks optimization (HHO) for feature selection, reducing redundancy and improving classification efficiency. Finally, XGBoost is utilized for classification, taking advantage of its strong generalization capabilities. <b>Results</b>: We evaluate our model on the King Abdulaziz University Mammogram Dataset, categorized based on BI-RADS classifications. Experimental results demonstrate the effectiveness of our approach, achieving 98.2% for accuracy, 98.0% for precision, 98.1% for recall, 98.0% for F1-score, 98.9% for the area under the curve (AUC), and 95% for the Matthews correlation coefficient (MCC), outperforming existing state-of-the-art models. <b>Conclusions</b>: These results validate the robustness of our fusion-based framework in improving breast cancer diagnosis and classification.https://www.mdpi.com/2075-4418/15/11/1361breast cancer classificationMAX-ViTgated attention fusion module (GAFM)Harris Hawks optimization (HHO)mammography analysis
spellingShingle Soaad Ahmed
Naira Elazab
Mostafa M. El-Gayar
Mohammed Elmogy
Yasser M. Fouda
Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer Classification
Diagnostics
breast cancer classification
MAX-ViT
gated attention fusion module (GAFM)
Harris Hawks optimization (HHO)
mammography analysis
title Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer Classification
title_full Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer Classification
title_fullStr Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer Classification
title_full_unstemmed Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer Classification
title_short Multi-Scale Vision Transformer with Optimized Feature Fusion for Mammographic Breast Cancer Classification
title_sort multi scale vision transformer with optimized feature fusion for mammographic breast cancer classification
topic breast cancer classification
MAX-ViT
gated attention fusion module (GAFM)
Harris Hawks optimization (HHO)
mammography analysis
url https://www.mdpi.com/2075-4418/15/11/1361
work_keys_str_mv AT soaadahmed multiscalevisiontransformerwithoptimizedfeaturefusionformammographicbreastcancerclassification
AT nairaelazab multiscalevisiontransformerwithoptimizedfeaturefusionformammographicbreastcancerclassification
AT mostafamelgayar multiscalevisiontransformerwithoptimizedfeaturefusionformammographicbreastcancerclassification
AT mohammedelmogy multiscalevisiontransformerwithoptimizedfeaturefusionformammographicbreastcancerclassification
AT yassermfouda multiscalevisiontransformerwithoptimizedfeaturefusionformammographicbreastcancerclassification