Explainable Transfer Learning Framework for Multi-Class Brain Tumor Classification from MRI Images

Authors

  • R. Yaswanth
  • Dr. R. venubabu

Keywords:

Efficientnetv2, Convolutional Neural Network (CNN), Brain Tumor Vision Transformers (ViT), Magnetic Resonance Imaging (MRI), Multilayer Perceptron (MLP).

Abstract

Brain tumor detection is a critical task in medical imaging, where accurate classification and localization of tumor regions in MRI scans directly influence diagnostic precision and treatment planning. In this work, the Brain Tumor MRI dataset is utilized for multi-class classification and detection tasks. For classification, advanced transfer learning architectures including EfficientNetV2, Vision Transformer (ViT-B16), DenseNet121, and Xception were employed, along with ensemble techniques based on average, weighted average, and geometric mean to enhance predictive reliability. For tumor localization and detection, multiple YOLO variants (YOLOv5, YOLOv8, YOLOv9, and YOLOv11) were implemented with bounding box annotations in YOLO format. Additionally, explainable AI techniques such as Grad-CAM were incorporated to generate heatmaps highlighting discriminative tumor regions that influence model predictions. Experimental results demonstrate that DenseNet121 and Xception models achieved the highest classification accuracy of 99.54%, outperforming other classifiers. For detection, YOLOv8 delivered superior performance with a mean average precision (mAP) of 95.1%. These findings validate the effectiveness of integrating classification and detection frameworks with interpretability methods for robust and reliable brain tumor diagnosis. Furthermore, a Flask-based web application was developed to deploy the trained models, enabling users to upload MRI images and visualize tumor localization, predicted class, and confidence scores through an interactive interface.

Downloads

Published

2026-11-16

How to Cite

Yaswanth, R., & venubabu, D. R. (2026). Explainable Transfer Learning Framework for Multi-Class Brain Tumor Classification from MRI Images. Adolescência E Saúde, 1035–1043. Retrieved from https://adolescenciaesaude.com/index.php/aes/article/view/3454

Issue

Section

Original Articles