Quickstart
Use GitHub Codespaces with the RunClientServerRecon.ipynb notebook to run a complete client/server reconstruction workflow.
GitHub Codespace is a cloud-hosted, web-oriented service that allows a complete development environment to be set up with a single click. It uses Docker images that are set up using the devcontainer.json file in this repo.
Start with GitHub Codespaces
A GitHub account is required to use GitHub Codespaces. Free personal usage includes 120 hours per month of compute time.
-
Open the repository in Codespaces
- Navigate to the python-ismrmrd-server repository on GitHub
- Click the green Code button → select Codespaces → click Create codespace on master
- Wait for the environment to initialize (typically <2 minutes)
- When complete, a web version of Visual Studio Code will be shown
- The environment is pre-configured and all required Python libraries are installed
-
Open RunClientServerRecon.ipynb
- In VS Code, open the file browser and navigate to
RunClientServerRecon.ipynbin the workspace root - Click on the notebook to open it
- In VS Code, open the file browser and navigate to
-
Execute the notebook cells
- The notebook contains the following workflow steps:
- Generate sample raw k-space data (Shepp-Logan phantom)
- Visualize the k-space data
- Start the MRD server (this is done in the Run and Debug tab of Visual Studio code, not within the notebook itself)
- Send data for reconstruction using the client
- Display the results
- Run each cell in order by clicking the play button or pressing
Ctrl+Enter - Output images and metadata are displayed in the output cells
- The notebook contains the following workflow steps:
-
View results
- Reconstructed images are displayed in the output cells
- Reconstruction metadata is also shown
- Generated data files are saved to the
data/directory
Next steps
- Explore the architecture documentation to understand the server components
- Implement an example custom module for image filtering
- See the environments section for alternative setup options (Conda, uv, Docker)