Deep Learning Based Brain Tumor Detection Using MRI Scans

This project aims to classify brain MRI scans into four categories: Normal, Glioma Tumor, Meningioma Tumor, and Pituitary Tumor, using advanced deep learning techniques. Leveraging a variety of powerful pretrained models, such as VGG19, DenseNet201, ResNet152V2, EfficientNetB5, and ConvNeXt-Base —the system is designed to explore and compare model performance on complex medical imaging data.

Key features include:

• Comprehensive preprocessing and augmentation of MRI images
• Implementation of multiple transfer learning architectures using TensorFlow and Keras
• Rigorous evaluation using accuracy, precision, recall, F1-score, and confusion matrix
• Visualization of model performance and predictions to support interpretability

This project provides a scalable and automated solution to assist radiologists in the early detection and classification of brain tumors from MRI scans.

Dataset

RGB Dataset

Greyscale Dataset