MediScanAI
A medical image analysis prototype with a microservice architecture and AI models. Not a clinical tool.
The problem
Integrating inference tasks of variable duration without blocking the rest of the system.
What I built
A Spring Cloud Gateway API Gateway routes to IAM, report, notification and administrative services. A FastAPI service with TensorFlow classifies, and an ML Service with MONAI explains and trains. React and Flutter frontends.
Outcome
An engineering prototype with many services; it is not presented as a validated clinical tool.
Decisions that mattered
- 01
A single entry point
The API Gateway concentrates access and delegates authentication to the IAM service.
- 02
One service per responsibility
IAM, reports, notifications and administration are Spring Boot services, each with its own database.
- 03
Slow work, kept apart
Inference and training run in separate Python services, and emails go out through RabbitMQ.
- 04
Explainable results
The ML Service adds explainability with Captum and an X-ray model built with MONAI.
Technologies
- Spring Cloud Gateway
- Spring Boot
- FastAPI
- TensorFlow
- MONAI
- RabbitMQ
- React
- Flutter
C4 model. The pulsing line is the main path of a request.
Level 1 · who uses the system and what it connects to
Level 2 · the applications and data stores it is made of
Level 3 · the components inside the API
- Person
- System
- Container
- Database
- External
- Main path
- 01Imaging study
- 02API Gateway
- 03AI Service
- 04ML Service
- 05Report
- 06Notice