Page 78 - 2024F

UEC Int’l Mini-Conference No.53 71

BTDCGANet: An Advanced Deep Transfer Learning and Generative

Adversarial Network- Based Data Augmentation Approach for Brain

Tumor Classification from Magnetic Resonance Imaging (MRI) Scans

2
Shabnur Anonna AKHY ∗1 , Hayaru SHOUNO , and Tahir HUSSAIN 2
1
UEC Exchange Study Program (JUSST Program)
2 Department of Informatics
The University of Electro-Communications, Tokyo, Japan

Keywords: Generative Adversarial Networks (GANs), Transfer Learning (TL), Deep Learning (DL),
MRI, Brain tumor classification, Explainable AI

Abstract
A brain tumor is an abnormal growth of cells in the brain caused by uncontrolled cell division.
Diagnosing brain tumors in magnetic resonance imaging (MRI) scans is challenging due to variations
in tumor structure and the complexity of image segmentation. To address these challenges, this study
proposes a deep learning-based approach that integrates EfficientNetB0 for feature extraction and
Deep Convolutional Generative Adversarial Networks (DCGANs) for data augmentation. Addition-
ally, a comprehensive evaluation of twelve transfer learning models, including GoogLeNet, ResNet34,
DenseNet102, MobileNetV2 and VGG-16, conducted for brain tumor classification. The expected ap-
proach enhances model accuracy and reliability by leveraging synthetic data generated through GANs.
To improve the interpretability of model predictions, Explainable AI techniques, specifically Local In-
terpretable Model-Agnostic Explanations (LIME), will be employed. The study analyzes 7,023 MRI
images and trains BTDCGANet on a dataset comprising gliomas, meningiomas, non-tumorous cases,
and pituitary tumors. Furthermore, this research explores a comparative analysis of GANs and diffusion
models to determine the most effective approach for early brain tumor detection using MRI images.

∗
The author is supported by JASSO Scholarship.