Precision System for Early Detection of Brain Tumor

This research presents an intelligent system for brain tumor detection using deep learning andexplainable AI techniques. The proposed framework employs a convolutional neural network (CNN)model trained on MRI scans to classify tumors with high accuracy, complemented by a web-based interfacefor clinical deployment. The system integrates gradient-based saliency mapping to provide visualexplanations of model predictions, enhancing interpretability for medical professionals. A Flask-basedREST API serves as the backend, processing image uploads and delivering real-time predictions along withconfidence scores. The implementation includes Power BI integration for analytics and monitoring ofdiagnostic outcomes. Experimental results demonstrate the system's effectiveness in tumor classification,with quantitative evaluation of both diagnostic accuracy and computational performance.The solution addresses key challenges in medical AI, including model transparency and clinical workflowintegration. The saliency maps offer radiologists intuitive visualizations of the model's decision-makingprocess, while the lightweight web architecture ensures practical deployment in healthcare settings. Thiswork contributes to the field of computer-aided diagnosis by combining state-of-the-art deep learning withexplainability features and scalable system design.