Title: AI-Powered brain cancer prediction web-application
Authors: Mahmudu J Mcha
Volume: 10
Issue: 5
Pages: 94-101
Publication Date: 2026/05/28
Abstract:
Brain cancer is a life-threatening neurological condition that requires early and accurate diagnosis to improve patient survival and treatment outcomes. In many low-resource countries, including Tanzania, the diagnosis of brain tumors is severely constrained by a critical shortage of radiologists and neurosurgeons, limited access to advanced diagnostic tools, and delays in clinical decision-making. This review presents Neuropredict, an AI-powered brain cancer prediction system designed to support healthcare professionals by automating the analysis of brain MRI scans using deep learning techniques; the manuscript describes planned methods, anticipated evaluations, and a roadmap for future validation. The system integrates a convolutional neural network (CNN) model within a full-stack web application to classify MRI images into binary outcomes indicating the presence or absence of brain cancer. The development of Neuropredict will follow the CRISP-DM methodology, encompassing data understanding, preprocessing, modeling, evaluation, and deployment. MRI datasets will be obtained from publicly available repositories; where permissible, these will be supplemented with locally sourced, de-identified MRI images from Tanzanian hospitals to enhance contextual relevance to enhance contextual relevance. Image preprocessing techniques such as resizing, normalization, and data augmentation will be applied to improve model robustness and generalization. The trained CNN model will be deployed through a FastAPI backend and a user-friendly web interface to enable near real-time clinical use in pilot settings. System validation will be conducted using external MRI scans to assess predictive performance, measuring confidence scores and agreement with traditional clinical assessments. If validated, AI-assisted diagnosis could reduce diagnostic time while maintaining high accuracy, thereby helping to alleviate the burden on limited medical specialists. Despite its promising performance, the system will initially focus on binary classification and may not support tumor grading or full DICOM/PACS integration until later development. Overall, this review highlights the potential of AI-driven diagnostic tools like Neuropredict to enhance early brain cancer detection, particularly in resource-constrained healthcare settings, while emphasizing the need for further validation, explainability, and large-scale clinical deployment.