NeuroScan AI: An Explainable Deep Learning and RAG-Based Clinical Decision Support System for Multi-Class Brain MRI Diagnosis

It is very important to find neurological disorders like brain tumours and Alzheimer's disease early so that patients can get the best care and treatment. Radiologists can manually analyse MRI scans, but this can take a long time and lead to different diagnoses. This project introduces NeuroScan AI, an advanced medical imaging platform that utilises deep learning methods to automatically analyse brain MRI scans. The system uses a Convolutional Neural Network (CNN) model to sort MRI images into three diagnostic groups: normal anatomy, brain tumour, and Alzheimer's disease. The system uses Grad-CAM visualisation to make the model's predictions easier to understand and more trustworthy. This shows the parts of the MRI scan that had the biggest impact on the neural network's decision. The platform is built on the Django framework, which makes it safe for users to log in, upload MRIs, and see their dashboards based on their roles as patients or doctors. Patients can upload MRI scans through the patient portal, and medical professionals can look at AI-generated predictions through an expert review interface. The system also has a Retrieval-Augmented Generation (RAG) module that gets relevant clinical knowledge and gives medical insights about the detected condition in context. The platform also supports automatic report generation and expert validation to help doctors make decisions. NeuroScan AI wants to make it easier to find neurological disorders in MRI scans by using deep learning, explainable AI, and medical knowledge retrieval.