EFT-SA-Net: An Attention-Enhanced EfficientNetB6 Framework for Accurate and Explainable Brain Tumour Classification from MRI Scans

Bagam, Ajay Kumar (2025) EFT-SA-Net: An Attention-Enhanced EfficientNetB6 Framework for Accurate and Explainable Brain Tumour Classification from MRI Scans. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (782kB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Brain tumor diagnosis remains an important clinical challenge due to the complexity and variability of MRI images. This study aims to develop an advanced, explainable deep learning framework that enhances both the accuracy and interpretability of tumor classification. The proposed EFT-SA-Net which is a EfficientNetB6-based hybrid model integrated with Spectral, Temporal and Self-Attention mechanisms, is designed to capture fine-grained spatial and contextual type of dependencies within MRI scans. The study has been used the Kaggle Brain Tumor MRI dataset which is undergoing preprocessing, SMOTE-based data balancing and model optimization using Adam and categorical cross-entropy. Experimental evaluations compared CNN, Xception, InceptionV3, EfficientNetB6, and EFT-SA-Net, with the proposed model achieving the highest accuracy of 98% showing superior results. Integration of Explainable AI (LIME) provided interpretability by visually showing tumor regions influencing the model’s decisions reinforcing medical trust and diagnostic transparency. The results confirm that EFT-SA-Net not only outperforms conventional deep learning models but also bridges the gap between automation and clinical interpretability. While computational complexity presents a limitation, the framework shows a major advancement toward AI-driven, reliable and explainable medical diagnostics with potential for real-world clinical deployment and future multi-modal medical imaging integration.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Qayum, Abdul UNSPECIFIED
Uncontrolled Keywords:	Brain Tumor Classification; Deep Learning; EfficientNetB6; Attention Mechanisms; Explainable Artificial Intelligence (XAI)
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence R Medicine > Healthcare Industry Q Science > Life sciences > Medical sciences > Pathology > Tumors
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	07 Sep 2026 09:27
Last Modified:	07 Sep 2026 09:27
URI:	https://norma.ncirl.ie/id/eprint/9849

Actions (login required)

View Item