3dslicer-claude-bridge by brainbloodbarrier - MCP Server

MCP Slicer Bridge

Model Context Protocol (MCP) server providing Claude Code with programmatic access to 3D Slicer for AI-assisted medical image analysis.

Overview

The MCP Slicer Bridge enables Claude Code to interact with 3D Slicer, a powerful open-source platform for medical image analysis, surgical planning, and radiomics research.

Use Cases:

AI-assisted surgical planning with real-time visualization
Automated radiomics analysis workflows
Interactive medical image segmentation
Teaching medical imaging concepts with AI guidance

Features

Tools (12)

Tool	Description
`capture_screenshot`	Capture PNG from any viewport (axial, sagittal, coronal, 3D, full)
`list_scene_nodes`	Inspect MRML scene structure and node metadata
`execute_python`	Execute Python code in Slicer's environment
`measure_volume`	Calculate segmentation volumes in mm³/mL
`list_sample_data`	Discover available sample datasets
`load_sample_data`	Load sample datasets (MRHead, CTChest, etc.)
`set_layout`	Configure viewer layout (FourUp, OneUp3D, etc.)
`import_dicom`	Import DICOM files from folder into database
`list_dicom_studies`	List all studies in DICOM database
`list_dicom_series`	List series within a DICOM study
`load_dicom_series`	Load DICOM series as volume
`run_brain_extraction`	Brain extraction/skull stripping (HD-BET or Swiss)

Resources (3)

Resource	Description
`slicer://scene`	Current MRML scene structure
`slicer://volumes`	Loaded volumes with metadata
`slicer://status`	Connection health status

Prerequisites

1. 3D Slicer

Download 3D Slicer 5.0+ and enable WebServer extension:

View → Extension Manager → Search "WebServer" → Install
Restart Slicer
Modules → Developer Tools → WebServer → Start Server

2. Python Environment

# Install uv package manager
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install dependencies
cd slicer-bridge
uv sync

Installation

Add to ~/.claude/mcp.json:

{
  "mcpServers": {
    "slicer-bridge": {
      "command": "uv",
      "args": ["--directory", "/path/to/slicer-bridge", "run", "slicer-mcp"],
      "env": {
        "SLICER_URL": "http://localhost:2016"
      }
    }
  }
}

Restart Claude Code to load the configuration.

Usage Examples

Scene inspection:

"What volumes are loaded in Slicer?"

Screenshot capture:

"Show me an axial view of the current volume"

Volume measurement:

"Calculate the volume of the tumor segmentation"

Custom analysis:

"Apply a Gaussian blur with sigma=2.0 to the current volume"

Troubleshooting

Error	Solution
`SLICER_CONNECTION_ERROR`	Start Slicer, enable WebServer, check port 2016
`SLICER_TIMEOUT`	Restart Slicer (may be frozen)
`PYTHON_EXECUTION_ERROR`	Add `import slicer` to code
MCP server not starting	Check `mcp.json` syntax, verify path, run `uv sync`

Development

uv run pytest -v                     # Run tests
uv run pytest --cov=slicer_mcp       # Coverage
uv run black src tests               # Format
uv run ruff check src tests          # Lint

Security Notice

This is an educational/research tool for localhost use only. Not suitable for clinical data or production environments. See for details.

brainbloodbarrier/3dslicer-claude-bridge