microsoft/uwisc-workshop-2026

microsoft/uwisc-workshop-2026

Language: Python

License: MIT

Stars: 4

Forks: 5

Open issues: 3

Created: 2026-05-14T07:41:08Z

Pushed: 2026-06-04T17:53:16Z

Default branch: main

Fork: no

Archived: no

README:

UW–Madison 2026 Workshop: Building Radiology Agents with State of the Art Foundation Models

This repository supports a hands-on workshop on building radiology AI agents in VS Code with GitHub Copilot. The workshop is organized into four chapters, each in its own folder under [chapters/](chapters/).

What you'll build

1. Chapter 1 — Simple agent + skill ([chapters/chapter1-simple-agent/](chapters/chapter1-simple-agent/)): use a general-purpose LLM with a prompt-based skill to draft structured radiology reports. 2. Chapter 2 — Domain MCP server ([chapters/chapter2-zero-shot/](chapters/chapter2-zero-shot/)): wrap Microsoft **MedImageInsight** zero-shot classification in a custom MCP server the agent can call. 3. Chapter 3 — Adapter training ([chapters/chapter3-adapter/](chapters/chapter3-adapter/)): train a small adapter on top of frozen MedImageInsight embeddings. 4. Chapter 4 — Patient context ([chapters/chapter4-knowledge-sources/](chapters/chapter4-knowledge-sources/)): ground agent outputs in a (synthetic) patient history.

Workshop layout

• [chapters/](chapters/) — one folder per chapter. Each has a README.md, a starter/ skeleton, and an answer-key/ reference.
• [samples/](samples/) — 30 sample chest X-rays (NIH 2017) you can hand to the agent.
• [eval-results/](eval-results/) — reference scoring numbers we ship for the part-to-part comparisons.

> The data and patient records in this repo are synthetic or de-identified and are not for clinical use. The images are from the 2017 NIH Kaggle dataset. Ground-truth report text in eval-results/ is model-generated from NIH labels for reference scoring only; it is not radiologist-written.

Prerequisites

Install and verify the following before the workshop. Wi-Fi and time will be limited on the day.

• VS Code (latest stable): https://code.visualstudio.com/
• GitHub Copilot and GitHub Copilot Chat extensions, signed in with an account that has Copilot access.
• Git, and a local clone of this repository opened as your VS Code workspace.
• Python 3.11+ on your PATH.
• uv (https://docs.astral.sh/uv/). Then run uv sync from the repo root to install all workshop dependencies.

MedImageInsight credentials

Chapters 2 and 3 hit a shared MedImageInsight endpoint. Credentials (MI2_MODEL_ENDPOINT and MI2_MODEL_API_KEY) are distributed at the start of the session. Paste them into chapters/chapter2-zero-shot/answer-key/mcp/medimageinsight-server/.env (start from the .env.example template in the same folder).

Quick check

code --version
git --version
python --version
uv --version

If any command above is missing or reports an unexpected version, resolve it before arriving.