Brain Tumor MRI Classification Using a Dual-Branch Framework of Lightweight CNN and VGG16 Transfer Learning

Abstract

Brain tumors are a leading cause of cancer-related mortality, and their early and accurate classification from magnetic resonance imaging (MRI) is critical for treatment planning. Manual interpretation is time-consuming and subject to inter-observer variability. This paper presents an automated four-class brain-tumor classification system that combines a lightweight convolutional neural network (CNN) with a fine-tuned VGG16 transfer-learning model for parallel inference on MRI images. The models distinguish glioma, meningioma, pituitary, and no-tumor cases. The trained models are integrated into a Flask-based web platform that returns the predicted class, the confidence distribution, and a side-by-side comparison of the two models. On the test set, the lightweight CNN achieves 96.19% accuracy while the VGG16 model reaches 97.97%, with VGG16 consistently outperforming the CNN across all four classes. The results show that transfer learning with a deep backbone provides a reliable and practical approach for computer-aided brain-tumor classification.