🧠 Table of Contents:

RAISO – Building a Radiology AI Second Opinion App on Azure

1. Intro: When Radiologists Need a Teammate, Not a Replacement

   Why I built RAISO, and the real-world problem it tries to help with.

2. What is RAISO?

   The concept, intended users, and how it fits into the clinical workflow.

3. Architecture Overview: Azure All the Way

   A high-level view of the stack: frontend, backend, AI, storage, and security.

4. Integrating the AI Model: The Second Opinion Brain

   • Model choice, training (briefly), and deployment
   • How inference is triggered and how results are shown to the user

5. Key Components & How They Work Together

   • Streamlit frontend
   • Azure services used (Azure Container App, Data lake Gen2, Azure Search AI, ..etc.)
   • AI model backend & inference
   • Pseudonymization and secure handling of clinical data

6. Privacy, Compliance & Data Ethics

   How I made RAISO HIPAA-ish (or GDPR-ish) friendly and secure on Azure

7. Lessons Learned & Developer Tips

   If I could tell Past Me anything before starting RAISO...

8. Get Involved / Try It Out

   Where to find the GitHub repo, how to deploy your own RAISO, and how to reach me

9. Closing Thoughts: AI as a Clinical Companion

   Can AI and radiologists be teammates? I think so. Here’s why.

🩻 1. Intro: When Radiologists Need a Teammate, Not a Replacement

Let’s be honest: “AI will replace doctors” is one of those headline-grabbing myths that just won’t go away. But anyone who’s actually worked in healthcare knows that medicine — especially radiology — isn’t a solo game, and it’s definitely not one that machines can play alone. What doctors need isn’t a replacement. What they need is a tireless, always-on teammate who doesn’t get tired of reading 50 chest X-rays in a row.

That’s where RAISO comes in — short for Radiology AI Second Opinion. It’s a web app I built to offer radiologists (and curious clinicians) a streamlined, AI-assisted second opinion tool for medical imaging. Think of it as that ultra-nerdy colleague who sees patterns in images and never takes a coffee break.

RAISO was born out of a simple but powerful question:

What if AI could give radiologists a second opinion on demand — one that’s fast, secure, and cloud-native?

In this post, I’ll walk you through how I built RAISO from scratch on Azure, what worked, what didn’t, and how I combined open-source tools, Azure services, and some clever engineering to bring it to life.

Spoiler alert: you can try the app yourself, and the frontend is open-source. Whether you’re a developer, a healthcare innovator, or someone who likes to tinker with AI in their spare time, there’s something in here for you.