Overview
MedScan AI brings multiple deep learning models together in a single web application, allowing users to upload scans, select the appropriate diagnostic model, and receive prediction results with confidence scores and downloadable reports.
The problem
Various research models have been developed for medical image diagnosis using MRI, X-ray, and CT scans, but these solutions are often limited to individual research implementations. There is a lack of a unified platform that provides users with access to multiple AI-based diagnostic models across different medical imaging modalities through a single application.
The solution
I built a full-stack platform that integrates four CNN models with a web interface for scan uploads, model selection, authentication, prediction results, confidence scores, and downloadable analysis reports.
Approach
The platform connects multiple CNN-based classification models with a full-stack application. Users upload a scan, select the appropriate model, and receive per-class prediction probabilities to understand the model output.
Key features
- Upload X-ray, MRI, and CT scans
- Select the appropriate diagnostic model
- JWT-based user authentication
- Per-class prediction probabilities and confidence scores
- Downloadable analysis reports
- Contact page for user inquiries
What I learned
Building MedScan AI strengthened my experience in integrating deep learning models with full-stack applications and presenting model predictions and confidence scores through a practical user interface.