Skip to main content

dicom-agent-mcp

An MCP (Model Context Protocol) server that gives AI agents one set of tools for both DIMSE archives (C-ECHO, C-FIND, C-GET, C-MOVE, C-STORE) and DICOMweb archives (QIDO-RS, WADO-RS, STOW-RS).

It also includes pixel and geometry tools that work on local files: rendering, statistics, measurement, geometry checks, multiplanar reconstruction (MPR) and de-identification. This is the tool layer for Topic 3 (Validated Tool Execution). The verification loop comes later; see ROADMAP.md.

Not for clinical use. This is a research scaffold. Use only with test or de-identified data.

Tools

Tool DIMSE node DICOMweb node Notes
list_nodes – – Lists configured nodes; secrets are never returned
echo C-ECHO GET /studies?limit=1 Connectivity check
search C-FIND QIDO-RS Levels: patient, study, series, instance. Supports wildcards and date ranges. Paging uses offset/limit. QIDO has no patient level
get_metadata C-FIND (indexed attributes only) WADO-RS /metadata Returns keyword-based JSON; bulk data is summarised
retrieve C-GET WADO-RS Saves files to storage_dir/<study>/<series>/<sop>.dcm
move C-MOVE – Destination is a node name or an AE title
store C-STORE STOW-RS Accepts files or directories inside the sandbox
transfer retrieve + store retrieve + store Works across protocols (e.g. PACS → DICOMweb)
list_local_files – – Lists cached or allowed DICOM files
read_local_metadata – – Reads a local file header without pixel data

Pixel and geometry tools (local files)

Tool What it does
render_image Renders one frame to a PNG and returns it alongside JSON. Windowing comes from explicit centre/width values, a preset (lung, mediastinum, soft_tissue, abdomen, liver, bone, brain, subdural, stroke), the file's window, or 0.5–99.5 percentiles. Corrects for non-square pixels and MONOCHROME1, and labels the edges with patient directions (for example R/L, A/P)
pixel_statistics Reports min, max, mean, standard deviation, percentiles and a voxel count for a frame or a rectangular/elliptical ROI. ROIs can be given in pixels or mm. Values are in modality units (HU for CT) and the ROI area is reported in mm²
measure_distance Measures the in-plane distance in mm between two pixels and returns both points in patient (LPS) coordinates
check_geometry Groups files into series and checks whether each one can be treated as a volume. Issues are reported with codes such as MISSING_SLICES, NON_UNIFORM_SPACING, DUPLICATE_POSITIONS, MIXED_ORIENTATION, OBLIQUE, SHEARED_STACK and INSTANCE_ORDER_MISMATCH
render_mpr Renders an axial, coronal or sagittal slice reformatted from a series that passes check_geometry. The PNG uses true mm aspect and radiological orientation
deidentify Writes de-identified copies under storage_dir/deidentified using a subset of the Basic Profile. See De-identification

Every pixel result includes an input_fingerprint: a SHA-256 hash of the decoded pixels and geometry. It does not change with the transfer syntax, and the planned verification loop will use it to bind results to their inputs.

Each tool carries MCP annotations (readOnlyHint, destructiveHint, idempotentHint) so a harness can decide which tools need approval.

Install

python -m venv .venv
.venv\Scripts\python -m pip install -e ".[dev]"
# Optional: decoders for JPEG, JPEG-LS, JPEG 2000 and RLE compressed pixel data
.venv\Scripts\python -m pip install -e ".[codecs]"

Configure

No config file is needed. Pass nodes and paths as arguments in the harness config; add a YAML file only for advanced settings.

Arguments

Argument Meaning
--node SPEC Adds a node. Repeatable. See the node spec below
--storage-dir DIR Local cache for retrieved files and renders. Default ~/.dicom-agent-mcp/cache
--allow-path DIR Extra directory that the local-file and pixel tools may read. Repeatable
--calling-ae AE AE title this server uses for DIMSE. Default DICOM_MCP; register it on your PACS
--config FILE Optional YAML file (see below)
--print-config Prints the resolved settings and exits. Useful for checking a harness setup

Node spec: name=URL[;option;option…]

URL Node type
dimse://AE_TITLE@host:port DIMSE (C-FIND, C-GET, C-MOVE, C-STORE)
http://… or https://… DICOMweb base URL (QIDO-RS, WADO-RS, STOW-RS)
Option Meaning
read-only store, move and transfer refuse to write to this node
token-env=VAR DICOMweb bearer token read from environment variable VAR
user=NAME;password-env=VAR DICOMweb basic auth; the password is read from VAR
no-verify-tls Skip TLS certificate checks (test servers only)
transfer-syntax=UID Transfer syntax requested from WADO-RS. Empty means the server default; the default * keeps the stored encoding

Examples:

--node orthanc=dimse://ORTHANC@localhost:4242
--node orthanc-web=http://localhost:8042/dicom-web
--node "azure=https://<workspace>-<service>.dicom.azurehealthcareapis.com/v2;token-env=AZURE_DICOM_TOKEN;read-only"

Secrets never go in arguments. Arguments are visible in process lists and are often committed in mcp.json. Put tokens and passwords in environment variables and pass only the variable name.

With no --node, the server still starts: the local-file and pixel tools work on files under the allowed paths.

Environment variables

Every argument has an environment-variable equivalent, which suits Docker and CI:

Variable Equivalent
DICOM_AGENT_MCP_NODES --node; several specs separated by spaces
DICOM_AGENT_MCP_STORAGE_DIR --storage-dir
DICOM_AGENT_MCP_ALLOWED_PATHS --allow-path; several paths separated by ; on Windows or : elsewhere
DICOM_AGENT_MCP_CALLING_AE --calling-ae
DICOM_AGENT_MCP_CONFIG --config

Optional YAML file (advanced)

Use a file for settings that have no argument: timeouts, max_results, max_volume_voxels, deid_key_env and node descriptions. See config.example.yaml. If ./dicom-agent-mcp.yaml exists in the working directory, it is loaded automatically. Relative paths in a file resolve against the file's directory; relative paths in arguments and environment variables resolve against the working directory.

Precedence: config file, then environment variables, then arguments. Later sources override single values, nodes are merged by name (a --node with the same name replaces the file's node), and allowed paths are combined.

Run

dicom-agent-mcp --node orthanc=dimse://ORTHANC@localhost:4242                                   # stdio
dicom-agent-mcp --node orthanc=dimse://ORTHANC@localhost:4242 --transport streamable-http --port 8000
dicom-agent-mcp --node orthanc=dimse://ORTHANC@localhost:4242 --print-config                    # check settings

VS Code (.vscode/mcp.json)

{
  "inputs": [
    { "id": "azureToken", "type": "promptString", "description": "Azure DICOM access token", "password": true }
  ],
  "servers": {
    "dicom": {
      "type": "stdio",
      "command": "${workspaceFolder}/.venv/Scripts/dicom-agent-mcp",
      "args": [
        "--node", "orthanc=dimse://ORTHANC@localhost:4242",
        "--node", "orthanc-web=http://localhost:8042/dicom-web",
        "--node", "azure=https://<workspace>-<service>.dicom.azurehealthcareapis.com/v2;token-env=AZURE_DICOM_TOKEN;read-only",
        "--storage-dir", "${workspaceFolder}/dicom-cache",
        "--allow-path", "${workspaceFolder}/inbox"
      ],
      "env": { "AZURE_DICOM_TOKEN": "${input:azureToken}" }
    }
  }
}

Claude Desktop / Copilot CLI / other harnesses

{
  "mcpServers": {
    "dicom": {
      "command": "dicom-agent-mcp",
      "args": [
        "--node", "orthanc=dimse://ORTHANC@localhost:4242",
        "--allow-path", "C:/data/dicom-inbox"
      ]
    }
  }
}

Use absolute paths here: these harnesses may start the server in any working directory. To use a file instead, pass "args": ["--config", "C:/path/to/dicom-agent-mcp.yaml"].

Local Orthanc for testing

docker-compose up -d   # DIMSE on 4242, DICOMweb on http://localhost:8042/dicom-web (compose file not yet tested)

Sample prompts

These prompts assume three nodes: orthanc (DIMSE), orthanc-web (DICOMweb) and azure-dicom, for example:

--node orthanc=dimse://ORTHANC@localhost:4242 --node orthanc-web=http://localhost:8042/dicom-web
--node "azure-dicom=https://<workspace>-<service>.dicom.azurehealthcareapis.com/v2;token-env=AZURE_DICOM_TOKEN"

The tool the agent is expected to call is shown after each prompt.

Nodes and connectivity

Prompt Tool
"Which DICOM nodes are configured, and which ones are read-only?" list_nodes
"Check whether orthanc and azure-dicom are reachable." echo

Search and metadata

Prompt Tool
"Find all CT studies on orthanc from the last 30 days." search (C-FIND)
"List the series in study 1.2.840… on orthanc-web, with modality and number of images." search (QIDO-RS)
"Find patients whose name starts with DOE^." search at patient level (DIMSE nodes only)
"Show the acquisition parameters for series 1.2.840…: slice thickness, kVp and pixel spacing." get_metadata

Moving data

Prompt Tool
"Download the chest CT series from orthanc into the local cache." retrieve
"Send study 1.2.840… from orthanc to the AE title WORKSTATION1." move (C-MOVE)
"Upload the files in ./inbox/case42 to orthanc-web." store (STOW-RS)
"Copy series 1.2.840… from orthanc to azure-dicom." transfer (works across protocols)

Local files

Prompt Tool
"What DICOM files do I have locally?" list_local_files
"Read the header of ./dicom-cache/…/x.dcm. Is it compressed? What is the photometric interpretation?" read_local_metadata

Pixels and geometry

Prompt Tool
"Show me slice 40 of that series in a lung window." render_image (preset)
"Render it again with centre 40 and width 400." render_image (explicit window)
"What is the mean HU in a 10 mm circle centred at row 256, column 300?" pixel_statistics (ellipse in mm)
"Give me the minimum, maximum and 95th percentile for the whole frame." pixel_statistics
"How far apart in mm are pixel (120, 200) and pixel (180, 260)?" measure_distance
"Can this series be used as a 3D volume? Report any missing slices or uneven spacing." check_geometry
"Show the middle coronal slice in a bone window." render_mpr
"Show sagittal slice 100." render_mpr

De-identification

Prompt Tool
"De-identify the downloaded study, shift the dates and drop private tags." deidentify
"De-identify it but keep the original dates and the series descriptions." deidentify (date_mode=keep, keep_descriptions)

Multi-step workflows

  • "Find the latest CT for patient MRN123 on orthanc, download it, check its geometry, and show me an axial and a coronal view in a soft-tissue window."
  • "Measure the mean HU of the liver at slice 60 (a 15 mm circle at row 300, column 180) and report the input fingerprint."
  • "Download study X from orthanc, de-identify it, and upload the de-identified copy to azure-dicom."

Prompts the server should refuse

Prompt Expected result
"Store these files to <a node marked read_only>." Refused because the node is read-only
"Read C:\Windows\System32\config." Refused because the path is outside the sandbox
"Render a coronal MPR of a series that has missing slices." Refused, with the geometry issues listed

Safety

  • Path sandbox. store, list_local_files, read_local_metadata and the pixel tools only accept paths under storage_dir or allowed_paths.
  • UID validation. UIDs are checked before they are used in file paths or URLs.
  • read_only nodes. store, move and transfer refuse to write to them.
  • Result caps. max_results limits results; truncation is reported with truncated: true.
  • Pixels in context only by request. Raw pixel arrays are never returned. render_image and render_mpr do send a downsampled PNG to the model; a copy is saved under storage_dir/renders.
  • Volume cap. render_mpr refuses series larger than max_volume_voxels, and refuses series that fail check_geometry.

De-identification

deidentify does the following:

  • removes or empties identifying attributes;
  • drops private tags and overlay or curve groups;
  • replaces PatientName and PatientID with pseudonyms;
  • remaps instance UIDs consistently;
  • by default, shifts dates by a per-patient offset of 1 to 730 days.

It records what it did in DeidentificationMethodCodeSequence.

To get the same pseudonyms and UIDs across runs, set a secret key in the environment variable named by deid_key_env (default DICOM_AGENT_MCP_DEID_KEY). Without a key, each call uses a random key and the result reports linkable: false.

It does not remove text burned into the pixels; affected modalities produce a warning. It is a research aid, not a certified de-identification tool.

Architecture

server.py  (FastMCP tools)  →  service.py  (routing, sandbox, paging)
                                   ├── dimse.py     (pynetdicom SCU)
                                   ├── dicomweb.py  (httpx QIDO/WADO/STOW)
                                   ├── imaging.py   (windowing, geometry, ROI stats, MPR, fingerprints)
                                   └── deid.py      (Basic Profile subset)

Tests

.venv\Scripts\python -m pytest -q

The tests run end to end against an in-process pynetdicom PACS (two AEs so C-MOVE can be tested) and a mocked DICOMweb server. They need no network. The pixel tests use a synthetic CT series with known HU values, spacing and positions.

Limitations

  • STOW-RS builds the whole request in memory, and WADO-RS reads the whole response into memory.
  • DIMSE get_metadata only returns attributes that the archive indexes for C-FIND.
  • retrieve uses C-GET. Archives that only support C-MOVE need move to a storage SCP.
  • No TLS for DIMSE yet.
  • Pixel tools have been tested on synthetic CT and pydicom sample files only. Compressed transfer syntaxes, enhanced multi-frame objects and real-world data are not yet tested.
  • MPR uses nearest-slice sampling and does not resample to isotropic voxels.
  • Pixel tools work on local files; use retrieve first.

Roadmap

See ROADMAP.md.

Metadata

Release files for dicom-agent-mcp 0.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for dicom-agent-mcp 0.1.0
File Size Uploaded Provenance Actions
dicom_agent_mcp-0.1.0.tar.gz 55.5 kB Without provenance Details for dicom_agent_mcp-0.1.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for dicom-agent-mcp 0.1.0
File Interpreter ABI Platform Actions
dicom_agent_mcp-0.1.0-py3-none-any.whl Python 3 none any Details for dicom_agent_mcp-0.1.0-py3-none-any.whl

Total release size: 102.5 kB

Release files / dicom_agent_mcp-0.1.0.tar.gz

Download URL dicom_agent_mcp-0.1.0.tar.gz
Size 55.5 kB
Tags Source
SHA-256 checksum
How to use checksums
1f63c56639fd815f82944aa17c4e6af603d9687f7895e21318d1f9e109b0d524
BLAKE2b-256 checksum
How to use checksums
884dbb9465ebcedae4016c9648b08a65ef084a2210bbe88c38ab99057615baee
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.10

Release files / dicom_agent_mcp-0.1.0-py3-none-any.whl

Download URL dicom_agent_mcp-0.1.0-py3-none-any.whl
Size 47.0 kB
Tags Python 3
SHA-256 checksum
How to use checksums
8717401347de01b05ba25c7879fb30a9156df5e794aac4c7c05018ab4f63bbee
BLAKE2b-256 checksum
How to use checksums
b878213ce43b40adf4a1b06f60b2adc9bdbeeed9ecbe8927b4e10b5befec9705
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.10

Security for dicom-agent-mcp 0.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.1.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page