RadAI — AI X-ray Diagnostic System

• 

  frontend

• 

  frontend

• 

  backend

• 

  MEDICAL AI DIAGNOSTIC REPORT

1- Inspiration

RadAI was inspired by the growing role of AI in healthcare and the difficulty students face when learning X-ray interpretation. I wanted to build a system that makes medical imaging easier to understand and more accessible, especially for beginners.

2- What it does

RadAI is an AI-powered learning system that:

Accepts an uploaded X-ray image Uses AI to analyze the image Generates a structured educational explanation Produces a downloadable PDF report

It focuses on learning and explanation, not medical diagnosis.

3- How we built it Backend: FastAPI AI Engine: Ollama + MedGemma 1.5 Frontend: HTML, CSS, JavaScript PDF Generation: FPDF

Flow: User → Upload Image → FastAPI → AI Model → Structured Output → PDF Report

4- Challenges we ran into AI model latency causing timeouts (504 errors) Handling large X-ray images efficiently Integrating vision AI with backend smoothly Ensuring structured and reliable AI output Managing synchronous vs asynchronous API calls

5- Accomplishments that we're proud of Built a fully working AI + backend system Successfully integrated local AI (no cloud) Generated structured medical-style reports Designed system for educational impact, not just output

6- What we learned How to integrate AI models into backend systems Importance of system design for long-running tasks Handling API errors and debugging real-world issues Structuring AI prompts for better results Building scalable backend architecture

7- What's next for RadAI — AI X-ray Diagnostic System Improve model accuracy with real datasets Add interactive AI chat for students Implement RAG with medical knowledge Highlight regions of interest in X-rays Optimize performance and reduce latency Add voice-based explanations

Built With

• css
• fastapi
• html5
• javascript
• medgemma
• ollama
• python