AI -Powered Brain Tumor Detection and Classification from MRI Scans using EfficientNetV2B0 and Grad- CAM Visualization

This paper introduces an AI system for brain tumor classification and detection from MRI scans using the EfficientNetV2B0 deep learning model and Grad-CAM for interpretability. The system classifies brain MRI images into five classes: glioma, meningioma, pituitary, general tumor, and no tumor. Using transfer learning, the model attains an aggregate accuracy of 97.35% on an extensive test dataset. Grad-CAM visualizations are indeed effective at emphasizing tumor-relevant features, which are consistent with clinical radiological characteristics and enhance trust in the model’s outputs. A web interface offers a user-friendly image upload process and an easy-to-read representation of the results. The study proves to have a strong, interpretable solution for the automatic classification of brain tumors with high potential for clinical diagnostic assistance.