NeuraScan is an AI-driven web application that detects brain tumors from MRI scans using Mask R-CNN, designed to support early and accessible diagnosis through a practical clinical workflow.
Problem to Solve
Build an end-to-end clinical support tool that improves early diagnosis accessibility by turning MRI analysis into a fast, accurate, and easy-to-use web application for real users.
Solution Approach
The work covered the complete product pipeline:
- Annotating and segmenting MRI images for high-quality instance segmentation training data.
- Training a Mask R-CNN model to identify and segment tumor regions.
- Integrating the model into a FastAPI backend for real-time inference.
- Designing the interface and connecting the model, backend services, and frontend experience.
Impact
Delivered an end-to-end system from data preparation to deployment with accurate and responsive segmentation output, reducing the gap between deep learning research and clinically usable tooling.