Cogniscan-Deep Learning Based Brain Tumor Classification

Abstract

CogniScan is a web-based application that employs deep learning techniques to classify brain tumors from magnetic resonance imaging (MRI) scans in an efficient and structured manner. Conventional tumor identification depends heavily on manual interpretation by medical professionals, which is time-consuming and may introduce inconsistencies, particularly when processing large datasets. The proposed system automates this process by allowing users to upload MRI images and receive accurate classification results.The application supports multiple categories, including glioma, meningioma, pituitary tumor, and no tumor, enabling comprehensive diagnostic assistance. A custom Convolutional Neural Network (CNN) model is used for feature extraction and classification. To improve performance, preprocessing steps such as resizing and normalization are applied to ensure consistency across input images.The system incorporates secure user authentication, allowing users to register and log in before accessing prediction features. A user-friendly interface built with Streamlit provides smooth interaction, enabling image uploads and visualization of results along with confidence scores. SQLite is used for secure storage of user credentials and authentication management. Additional features include efficient data handling, rapid prediction generation, and reliable output delivery.Overall, the proposed system offers a scalable and accessible solution for automated medical image analysis, reducing manual effort and supporting healthcare professionals in diagnostic decision-making.