NeuroVisionNet: A Custom Transfer Learning with VGG16 for Brain Tumor Recognition

NeuroVisionNet: A Custom Transfer Learning with VGG16 for Brain Tumor Recognition

Aishwarya S Gamanagatti                     Shruthi Bagi                                      Vaishnavi G H

Dept. ECE                                               Dept. ECE                                            Dept. ECE

KLS Vishwanatharao Deshpande          KLS Vishwanatharao Deshpande      KLS Vishwanatharao Deshpande

Institute of Technology,                         Institute of Technology,                     Institute of Technology,

Haliyal 581329,India                              Haliyal 581329,India                    Haliyal 581329,India

aishu1647@gmail.com                          shruthibagi15@gmail.com                vaishnaviharchilkar@gmail.com

Prof. Suraj Kadli

Supervising Author, Dept. of Electronics and

Communication Engineering

KLS Vishwanatharao Deshpande Institute of Technology,

Haliyal 581329,India

sbk@klsvdit.edu.in

Abstract— Brain tumors are among the most Life-threating and complicated diseases to improve treatment outcomes. Conventional diagnostic approaches radiologists’ expertise. This research introduces an automated framework for detecting brain tumors, utilizing the capabilities of VGG16, a predefined convolutional neural network model, variability and delays in decision-making, fine-tuned via transfer learning on MRI brain scan images. Our methodology emphasizes efficient data preprocessing, robust feature extraction, and a lightweight classification head tailored for binary classification. Experimental results reveal exceptional classification accuracy, validating the model's applicability in prediction effectiveness. This research highlights the transformative power of deep learning in augmenting diagnostic workflows and enabling scalable, real-time medical support.

Keywords- Brain tumor, Deep learning, VGG16, MRI, Image based neural network, Cross - domain learning, Biomedical image analysis.