XyDromatics VNA Research

Loading…

Select a .lic file issued by the Synthology License Generator. The file is staged, decrypted, and — on successful parse — replaces the active license. Feature gating is re-evaluated on the next request; no restart is required.

Current engine

—

Configured

—

Changing the archive catalog DB engine or connection string requires a service restart. The SCP will refuse to start until the catalog is reachable.

Current backend

—

S3 secret

—

Azure conn string

—

The archive’s byte store. Local = filesystem under the data directory. S3 = an S3 or S3-compatible object store (AWS S3, Cloudflare R2, Backblaze B2, Wasabi, MinIO) — set an Endpoint for anything other than AWS S3. AzureBlob = Azure Blob Storage; prefer managed identity (no keys on disk) with the account URL, else a connection string. Changing the backend requires a service restart and does not migrate existing objects. Leave a Secret / Connection string blank to keep the stored (sealed) value. Note: on an object-store backend the local-filesystem maintenance passes (integrity bit-rot scan, on-disk OAS rewrite) are skipped — object durability is delegated to the store.

…

…

…

No search executed yet — fill in any field (or none, to list everything) and click Search.

Upload

Queue a server-side folder for import. The background worker walks the folder recursively, parses each .dcm file, and ingests through the same de-identifying path as C-STORE — every imported study is anonymized on ingest. License: batch_import. RBAC: manage_document_store.

Export up to 10 studies from the de-identified research archive as a single zip (synchronous streaming). For larger batches, use the cohort-driven Research Export workflow. License: export. RBAC: manage_document_store.

Content-based deduplication scan — scans up to N instances, computes SHA-256, identifies duplicates and reports potential savings. License: dedup. RBAC: view_storage_tiers.

Configure how long archived studies are kept. Policies compute matching study counts (preview); destructive execution is a separate, confirm-token-gated operator action and never runs unattended on this host. License: retention. RBAC: view_retention / manage_retention.

Lock listed Study Instance UIDs against deletion under retention policies. Active holds override retention sweeps. License: legal_holds. RBAC: view_legal_holds / manage_legal_holds.

Fixed hot / warm / cold tier catalog with per-study tier assignment. Movement is label-only in v1 — assigning a study to a tier records a real, queryable tier LABEL; the physical byte relocation across storage back-ends is a documented follow-on (the response reports bytes_relocated=false so nothing over-claims). License: storage_tiers. RBAC: view_storage_tiers (read) / manage_storage_tiers (move).

Look up a study's tier

Move a study

Walks the archive catalog + checks each instance's file is present and readable by fo-dicom. Records anomalies as findings; never deletes. License: integrity. RBAC: view_storage_tiers (read) / manage_storage_tiers (scrub).

Consolidate subject records across ID changes. Merging re-keys every archived study from source to target subject_id in one transaction. Un-merge is not available. License: patient_merge. RBAC: view_patient_merge / manage_patient_merge.

Merge audit log

IHE Object Change Management. Mark a study, series, or instance as REJECTED with a reason code; the archive hides active rejections from query. License: iocm. RBAC: view_iocm / manage_iocm.

Define a cohort

Coded-finding / body-part / age-band criteria match against the PHI-free concept store harvested from de-identified reports. The Impression text phrase (Tier 2A) additionally narrows the cohort against the sealed, de-identified Impression narrative — the text itself never leaves the host; only cohort membership is affected. Queries matching fewer than the k-anonymity floor of subjects are suppressed.

Internal Impression-Text Search (Tier 2A)

Snippets are egress-scrubbed excerpts of the de-identified Impression narrative, shown only when the match meets the k-anonymity floor. This is an INTERNAL search to help assemble cohorts — the retained text is never exported.

Cohorts

Deliver a cohort snapshot

Delivers the studies in a cohort's latest snapshot (already de-identified at ingest) to a destination. Large deliveries are resumable and throttled. Set a recipient to make it a billed, per-vendor delivery — each recipient gets an unlinkable, re-pseudonymized copy; leave it blank for an internal export.

Deliveries Refresh

A resumable, throttled delivery pipeline. Progress updates live; use Retry failed to re-run any dead-lettered studies (each study is billed at most once).

Assign a cohort to recipients

Deliver a cohort's latest snapshot to one or more saved recipients at once — or over time, by assigning again later. Each recipient gets an unlinkable, re-pseudonymized copy on its configured transport. Manage recipients on the Recipients page.

New recipient

Saved recipients Refresh

Catalogs DICOM objects that already exist in your configured storage backend — no bytes are copied and nothing is written to your lake. Indexed objects are recorded as not owned by this archive, so deleting a study removes only the catalog entry; your objects are never deleted.

Step 1 — Sample review

Reads a sample of objects and reports what it finds. Indexes nothing. Every object is checked for direct identifiers (birth date, address, accession, physician, institution…). What the check cannot decide for you is whether the patient names and IDs are anonymous — a de-identified object is required to carry a replacement name, so RESEARCH and SMITH^JOHN look identical to a tag scan. That is why you review the distinct values below before any index runs.

Step 2 — Review + attest, then index

Open a completed sample below, read the distinct patient names and IDs, and attest only if they are anonymous. Your attestation is recorded with your username and the time; it authorizes a full index of that prefix only.

Sample findings

Jobs Refresh

This deployment is licensed on a per-study plan. The figure below is the cumulative count of studies your archive has delivered to Downstream Recipients — the exact quantity Synthology bills on. Review it here to reconcile before your invoice. The per-study rate is set in your order and is not shown on this instance.

Read-only archive quality analysis over the catalog metadata — missing required tags, inconsistent demographics, non-standard modalities, orphaned instances. A scan is read-only over the archive (it refreshes the finding cache only); it never interprets images and carries no clinical function. Normalization is preview-only (a dry-run count of affected series). License: data_curation. RBAC: view_curation (read / scan) / manage_curation (ack / normalize).

Normalize modality (preview only)

IHE TCE — export a source study as a de-identified teaching-file bundle. The background worker de-identifies every instance (DICOM PS3.15 Basic Profile + burned-in-annotation reject; UIDs re-minted, non-reversible) and stages the bundle for download as a .zip. The download is the de-identified copy only — never source PHI. License: teaching_file. RBAC: view_document_store (read) / manage_document_store (create / delete).

Create export

Export jobs

Replicate de-identified archive instances to one or more peer VNAs via DICOMweb STOW-RS. The replication service polls for new instances and POSTs them to each enabled peer. License: replication. RBAC: view_replication / manage_replication.

Add peer

Configured peers

Recent jobs

Read-only FHIR R4 endpoints projecting the de-identified archive catalog into ImagingStudy + DiagnosticReport resources (standard FHIR JSON Bundle responses). License: fhir. RBAC: view_dicomweb.

DICOM Storage Commitment (PS3.4 Annex J) journal — records each N-ACTION-RQ / N-EVENT-REPORT-RQ pair and whether the SCP confirmed each requested SOP UID. License: storage_commitments. RBAC: view_storage_commit / manage_storage_commit.

AI-vendor-driven ingestion of SR, SC, and SEG DICOM objects. Each vendor authenticates with an opaque API key (X-Ai-Vendor-Key header); ingest enforces a per-vendor allowed-SOP-class list and de-identifies on write. License: ai_pipeline. RBAC: view_study_browser.

Register vendor

Ingest audit log

Read-only image preview of the de-identified corpus. Pick a study, browse instances, adjust window/level. This is a basic viewer; advanced viewing is out of scope for the research archive. License: diagnostic_viewer. RBAC: view_study_browser.

Per-profile config for the 5 IHE profiles. Each profile has its own license feature. Status: enabled = licensed + on, disabled = licensed not on, unlicensed = feature not in license. RBAC: view_dicomweb (read) / edit_dicomweb (save).

Runtime honesty: only ARI has live runtime today — it is served by the existing DICOMweb QIDO/WADO surface. The other four profiles (Dose Monitoring, SR Viewer, PDQm, SMART) persist config now; their protocol runtime is a per-profile follow-on and is not yet active.

VNA Research is RUO — Research Use Only.

Bulk-migrate archived studies out to another DICOM destination. The shared controller is mounted, but the outbound job/destination stores are intentionally not registered on this store-less research archive — enabling study egress is an operator/governance decision, not a UI toggle. License: migration. RBAC: view_destinations / edit_destinations.

Schedule recurring sends of matched studies to a destination. The shared controller is mounted, but the outbound rule/destination stores are intentionally not registered on this store-less research archive. License: scheduled_send. RBAC: edit_destinations.

Rule-based real-time fan-out of incoming studies to destinations. The shared controller is mounted, but the outbound rule/destination stores are intentionally not registered on this store-less research archive. License: smart_routing. RBAC: edit_destinations.