IronDICOM - KomvOS Features

Routing and rules

Four rule modes: Store and Forward, Direct (live, no queue), Review (held for a person) and Workflow.
Nested All-of and Any-of condition trees on DICOM attributes.
Rule priority by list position, with Stop on Match.
Custom Conditions for site-specific logic.
Load-balanced destination groups: round robin, or sticky by study or accession number.
Failover destinations that take over when the first choice is down.
Loop prevention.
Routing Simulator: test a rule against a real or sample object before it goes live.
Unmatched objects view, with a draft rule one click away.
Study Trace: everything recorded about one study, in order.
Priors Pre-fetch with four date-window modes.
Flows: scheduled and study-triggered workflows with DICOM and HL7 steps.

Storage SCP and SCU for every Storage SOP class.
Verification (C-ECHO), with a test button on every destination.
Query and retrieve: C-FIND, C-MOVE and C-GET, passed through or answered from local data.
Modality Worklist, federated across sources or answered locally from HL7 orders.
Storage Commitment.
MPPS relay.
One association per study for efficient sending.
A calling AE title you choose per destination.
A published DICOM Conformance Statement.

Per-destination transcoding to eleven target syntaxes: uncompressed, RLE, JPEG Lossless, JPEG-LS, JPEG 2000 lossless and lossy, and lossy JPEG.
Lossless output is checked sample for sample before it is accepted.
Lossy output is marked as lossy in the object, as the standard requires.
Codecs run in an isolated worker process, so a damaged image cannot take the router down.

HL7 v2 listeners over MLLP, each with its own size, time and connection limits.
A content-based HL7 rules engine.
HL7 destinations rendered from templates; message bodies are relayed byte for byte.
HL7 orders feed a local Modality Worklist.
DICOMweb: STOW-RS, QIDO-RS and WADO-RS.
FHIR R4: ImagingStudy, and Structured Report to Observation.
S3-compatible object storage destinations.
Folder destinations, written durably.

Reusable transformation profiles, applied per destination to a copy made at send time.
Tag actions: overwrite, delete, copy, regular-expression replace, new UID, consistent UID remap, consistent pseudonym, date shift, and reversible encrypt and decrypt.
Pixel region redaction with a click-to-draw tool, and watermarking.
Fail-closed: an object that cannot be redacted or de-identified is not sent.
Ingest profiles normalise headers on receipt and record the change in Original Attributes.
Leftover-identifier check, profile preview and an As-sent view of exactly what a destination received.
Body Part Mapping to bring order to free-text values.

Durable receive: an acknowledged object is on disk before the sender is answered.
An object is queued and audited together, or not at all.
Retry with back-off, tuned per destination, with a circuit breaker and a dead-letter list.
An unreachable destination is paused and probed, not failed.
Delivery verification.
Streaming receive and send within a fixed memory budget, for very large studies.
A spool laid out for millions of objects, on a volume or network share of your choice.
Spool retention you set, automatic configuration snapshots, and full backup and restore.

Two-box high availability: an Active and a Standby server that keep each other current, with no shared disk and no external cluster to run.
Automatic takeover in seconds when the Active server fails.
A floating IP address, so modalities and destinations never need reconfiguring.
An optional witness, either a file on a share or a third KomvOS, to settle which server is in charge.
Planned switchover by agreement, so patching does not interrupt imaging.
Split-brain protection: a server that cannot confirm it is alone holds, rather than guesses.
Active and Passive servers on shared PostgreSQL and shared storage are supported too.

Home dashboard and Live Status page.
E-mail alerts, SNMP traps and a health endpoint for your monitoring tools.
Failure summaries with plain-language hints about the likely cause.
Alert Advisor: suggests alert thresholds from your own traffic, using plain statistics.
Digest reports and an Activity report.
Database and memory sizing check.
Support Bundle for troubleshooting, containing no patient data.
Optional Hazon Cloud status feed: outbound only, counts only, off by default.

Everything is done in a web browser. Nothing to install on an administrator's PC, and nothing loaded from the internet.
Jobs grid: resend, copy-and-resend, header view and edit, Study Instance UID fix and batch actions.
Review and Sign with a wrong-patient check.
Batch edit, DICOM upload and Check image.
Two administrators cannot silently overwrite each other's changes.
Configuration backup and restore, plus automatic snapshots to roll back to.
A break-glass desktop console for emergencies.

Roles: Admin, Operator and Auditor.
Sign-in with local accounts, Active Directory or LDAP, or OpenID Connect single sign-on.
Authenticator-app codes (TOTP) for local accounts.
Idle sign-out, and a random one-time password at first start.
TLS 1.2 or later, or TLS 1.3 only, chosen per DICOM and HL7 connection. HTTPS for administration.
Spool encrypted with AES-GCM, with versioned keys.
Hash-chained audit log with chain verification and sealed archives.
Syslog export over UDP, TCP or TLS in RFC 5424 format.
Hardened DICOM parsing and listeners.
No telemetry and no vendor remote-access channel.

Optional AI Assistant that drafts routing rules and regular expressions for an administrator to review.
Off by default. No model is bundled, and none is contacted over the internet.
The Assistant never receives patient data.
Any feature that would read patient data is locked off at the server, cannot be enabled from a screen, and only ever talks to a model on the same machine.

Windows Server: a signed installer that runs KomvOS as a Windows service, with silent-install options for SCCM and Intune.
Linux: an official Docker image.
PostgreSQL: bundled for a single server, or use your own.
Fully air-gapped operation.
Licensed per server with a signed license file.
A 30-day grace period on expiry or a hardware change, with a banner and an e-mail, never a sudden stop.
Recovery licenses for restores and replacements.

Ask for a guided demonstration built around your modalities, your PACS and your routing rules.

Request a demo

Page updated

Google Sites

Report abuse