NeuroVision: An Explainable Deep Learning System for Brain Tumor Detection from MRI Images

Early detection of brain tumors is crucial for effective treatment planning and improved patient survival. However, manual interpretation of Magnetic Resonance Imaging scans is time consuming and subject to diagnostic inconsistencies. This paper presents NeuroVision, an explainable deep learning system for automated brain tumor detection and classification from MRI images. The proposed system employs a two stage architecture in which a Convolutional Neural Network first validates whether the uploaded image is a brain MRI, followed by an EfficientNetB0 model that classifies the scan into Glioma, Meningioma, Pituitary tumor, or No Tumor. To enhance model interpretability, Gradient weighted Class Activation Mapping is incorporated to localize the regions influencing the prediction. The system is deployed as a Flask based web application that enables image upload, result visualization, and automated report generation. Experimental evaluation demonstrates an overall classification accuracy of 94 percent with reliable performance across all tumor classes. The proposed system provides an accurate, efficient, and interpretable solution for computer aided brain tumor diagnosis.