Trusted-network DICOM workstation
One application, one tab per job: Dicommunication tests connectivity (ping, C-ECHO, C-STORE, C-FIND, MWL, HL7), Dicomtag Analytics queries and exports tags — including Philips Vue PACS — Dicom Anonymizer strips or scrambles PHI, Dicom Router watches a PACS and forwards new studies on a schedule, and Dicom Cleaner redacts burned-in pixel data. One shared config. Docker, Windows MSI, or macOS DMG — no Python, no venv, no vendor toolkit.
The name is a small grammar joke. DICOM plus communication share that extra m, so they become dicommunication.
Hospital PACS/VNA work stalls on the same questions: can we reach the port, does verification work, will storage accept an instance, does Query/Retrieve find studies, does the RIS offer a worklist, what does a study actually contain, and does anything need to be caught automatically or cleaned up before it moves on? Dicommunication answers those from one application. A tab row across the top switches between five tools — Dicommunication, Dicomtag Analytics, Dicom Anonymizer, Dicom Router, and Dicom Cleaner — and a shared Configuration tab holds the local AE, remote nodes, virtual identities, and logs once for all of them. A successful C-ECHO only proves Verification — the Testbench shows which SOP Classes were accepted or rejected so Storage, Q/R, and MWL stay distinct.
A tab row across the top switches products. Configuration — local AE, remote nodes, virtual identities, logs — is set once and shared by all five. The Windows MSI and macOS DMG install a single shortcut for the whole thing.
Test and analyze DICOM traffic
The network / DIMSE / HL7 workstation. Testbench, C-ECHO board, PING, C-STORE, PDF to Encapsulated PDF Storage, Study Root C-FIND, Modality Worklist C-FIND, a local worklist SCP, and HL7 v2 send over MLLP — with the association log and accepted vs rejected SOP Classes.
DICOM tag analytics
Study Root C-FIND at Study, Series, or Image level, including Philips Vue ELSCINT1 private keys. List and retrieve DICOM Structured Reports — Findings and Impression flattened from the Content Sequence, no language model — then export as copy, CSV, or JSON.
Anonymization
Query, retrieve, and de-identify studies, series, or images. Nuke, Fuzz, remove-patient-information, or per-tag Custom mode, with UIDs remapped consistently so a multi-instance study stays coherent afterward.
Automated routing
Scheduled Study Root C-FIND rules that watch a PACS on an interval or daily schedule. Add destination nodes and it auto-retrieves and forwards each new match — Start, Pause, or Stop any rule independently, with a live status dot and a full run history.
Pixel-level redaction
Query, retrieve, and black out a rectangle of burned-in pixel data — drag the region right on the retrieved image — then C-STORE the cleaned instance back out. Handles PHI baked into pixels, which tag scrubbing alone can't touch — a common fix for ultrasound studies, where patient demographics are often burned into the frame, especially after a patient mismatch needs correcting.
Philips Vue's private ELSCINT1 tags don't show up in generic DICOM tools. Dicomtag Analytics queries them directly, retrieves report text without opening Vue, and exports everything for spreadsheet-level analytics.
Collapsed "Vue PACS" key lists expose Tamar study and series tags — like Assign To Doctor, a confirmed matching key on Vue 12.2.8 — alongside standard DICOM keywords, as optional return or match fields.
Series keys unlock once a Study Instance UID is set; Image keys unlock once Study and Series UIDs are set. MR-only keys such as Repetition Time stay locked unless Modality is MR — click a result row to copy parent UIDs straight into the next level.
List SR reports finds every report series for a batch of studies. Retrieve report text C-MOVEs each one on its own association — Vue only accepts one C-MOVE per association — then flattens the Content Sequence into Findings, Impression, and full report text. No language model.
Count patients by name buckets a query by first letter of family name, starting from an approximate count and quietly re-running uncapped (up to 50,000 rows) for an exact one. Every table exports as copy (tab-separated for Excel), CSV, or JSON.
Linux, Windows, or macOS — pick the runtime that already includes everything. Locked-down PACS PCs should not get a Python from python.org.
Linux · Mac · Windows
docker compose up — image on GHCR. No local Python. Config lives in ~/.dicommunication.
Open the repoWindows 10 / 11
One Dicommunication shortcut — every other tool lives inside it as a tab. Own window (Edge WebView2), not a browser tab. Config in %LOCALAPPDATA%\dicommunication.
Download MSIApple Silicon
Drag Dicommunication.app to Applications — every tool lives inside it as a tab. Own window (WKWebView), not a Safari tab. Config in ~/.dicommunication.
Download DMGClone the repo and docker compose up, or grab the MSI / DMG from GitHub Releases. Open Dicommunication, add a node, press C-ECHO — then switch tabs for Dicomtag Analytics, Dicom Anonymizer, Dicom Router, or Dicom Cleaner.
Trusted-network lab workstation. Not a PACS, not a diagnostic viewer, not DICOM TLS, not a CE-marked medical device. Prefer synthetic data.