scientific-ai-omezarr-tutorial

This MCP server hosts tools to connect AI workflows with OME-Zarr conversions and related bioimage data processing. It exposes capabilities for converting images to the OME-Zarr format, validating datasets for interoperability, and integrating with AI analysis pipelines via the Model Context Protocol (MCP). When you start the server (via the Docker image), you can use the MCP tooling to enqueue conversion tasks, run batch processing, and orchestrate validation and optimization steps across large image collections. The server is designed to work with common bioimage formats and to provide cloud-friendly, scalable access to OME-Zarr outputs that can be consumed by downstream AI tools and analysis pipelines. Typical usage involves submitting a project of images, selecting a target output (OME-Zarr), and then monitoring progress and results through the MCP interface or API.

Recommended workflow:

• Prepare a collection of bioimage files (NRRD, TIFF, HDF5, etc.).
• Start the MCP server container and ensure it is reachable by your agent or client tooling.
• Use the server’s tools to convert to OME-Zarr, validate integrity, and apply any optimization (chunking, compression).
• Integrate the resulting OME-Zarr datasets into your AI pipelines via MCP tool selection, enabling automated analysis, model inference, or batch processing across datasets.

Prerequisites:

• Docker installed and running on your system (Docker Desktop on Windows/macOS or Docker Engine on Linux).
• Sufficient CPU/RAM and disk space to handle bioimage conversion and OME-Zarr generation.

Installation steps:

1. Pull and run the MCP server image:

# Start the MCP server container (detached) exposing port 8080
docker run -d --name scientific-ai-omezarr-mcp -p 8080:8080 \
  -e MCP_LOG_LEVEL=info \
  -e MCP_PORT=8080 \
  fideus-labs/scientific-ai-omezarr-tutorial-mcp

2. Verify the server is running:

docker ps | grep scientific-ai-omezarr-mcp

3. Optionally configure a client to connect to the MCP server API at http://localhost:8080 (adjust if using a remote host).

4. For persistent setups or production, consider docker-compose or Kubernetes manifests provided with the MCP image (adjust paths and environment variables as needed).

Tips and common notes:

• Environment variables: MCP_LOG_LEVEL controls verbosity (e.g., debug, info, warn, error). MCP_PORT should match the port exposed by the container if you access it from outside.
• Large datasets may require increased memory and longer processing time; monitor resource usage and adjust container limits accordingly.
• If you encounter connection issues, ensure the container port is exposed and that your client is pointed at the correct host/IP and port.
• This MCP server focuses on ngff-zarr related workflows (conversion to OME-Zarr, validation, optimization) and integration with AI toolchains via MCP.