Transfer Learning for Brain Tumor MRI Classification Using VGG16: A Comparative and Explainable AI Approach

Brain tumor detection through magnetic resonance imaging (MRI) is a complex investigation to conduct. Developing a fast and reliable clinical decision-making tool is paramount. Modern techniques like deep learning and convolutional neural networks (CNNs) have demonstrated great promise in automating the process of detecting tumor masses from MRI scans. In this study, we take a different approach by training a VGG16-based CNN, and instead of relying on single source dataset or black-box predictions, we merge two publicly available datasets (Figshare and Kaggle), introducing inter-dataset variability that simulates real-world diagnostic conditions. We start by preprocessing the data, use stratified splitting for training, testing and validation, and at last, we use data augmentation techniques; our model achieves a validation accuracy of 84.4% and demonstrates consistent performance across tumor types. Grad-CAM heatmaps highlight tumor regions with reasonable precision, even in some misclassified cases, thereby enhancing model transparency and trust. This work highlights the effectiveness of a lightweight, generalizable CNN architecture along with visual interpretability.