Technical Articles

Ultrasound DICOM Duplicate Instance Resend: Triage Order

GEPROBE
9 min read
68 views
#ultrasound DICOM duplicate resend#duplicate SOP instance UID#PACS duplicate study after resend#DICOM storage commitment#ultrasound probe repair
Free technical review

Same symptom on your system?

Tell us your model and what it is doing. A GEPROBE engineer replies with a diagnosis path, part options and pricing — usually within 6 hours, with no obligation.

Chat on WhatsApp
Prefer email?Email us
Guaranteed response within 6 hours

Your details are used only to answer your question — never shared or added to a mailing list.

Ultrasound DICOM Duplicate Instance Resend: Triage Order

An ultrasound DICOM duplicate instance resend is a display event with three possible owners, and the screen does not say which one. The cause may sit in the archive, in the retry, or in the viewer.

This guide orders the checks by elapsed time rather than theory, for HTM teams and field engineers who must decide within the hour.

Order the Checks by the Clock

A duplicate entry does not identify the failed layer. It says something changed in the object, the transport, or the presentation.

Work the ladder in order and stop at the first row that yields a conclusion. Each row carries a time box, so the triage ends even with missing evidence.

Step Time box Who Look at Conclusion
Symptom class 0–5 min Clinical user with HTM Study count, series count, tiles, banner text Two studies, extra series, extra tiles, unchanged
Object identity 5–20 min HTM with console access SOP Class UID (0008,0016) and SOP Instance UID (0008,0018) Same object or different object
Hierarchy 5–20 min Same Study Instance UID (0020,000D), Series Instance UID (0020,000E) Regrouped or regenerated
What was kept 20–45 min PACS administrator Stored object list, policy in force at event time A repeat exists in storage, or not
Why it re-sent 20–45 min IT with modality owner Queue entry, association result, trigger Transport retry or fresh export

Rows two and three are the ones teams skip. A study count is a presentation fact; a UID pair is an identity fact.

Three Surfaces, Three Different Duplicates

The same incident looks different depending on where you stand. Treat each surface as a witness with a narrow vocabulary.

Surface Repeat looks like Can prove Cannot prove
PACS web viewer Two worklist rows, or one row with a series count of two The query returned two result sets Whether one object was stored twice
Diagnostic workstation Duplicate thumbnails, a second cine, a second report draft The local cache holds two objects Whether the archive holds one or two
Scanner send queue "Sent", "pending", "failed" badges What the sender believes about delivery What the receiver retained

A query result and a stored object are different claims. The viewer groups what the archive returns and cannot say why.

Count Before You Resend

Put numbers on the suspicion before anyone retries. Two counts and three averages are enough.

Let A be attempts including automatic retries, n̄ the mean objects per attempt, U the distinct SOP Instance UIDs across all attempts, B̄ the mean object size in megabytes, and t the viewer load time per object.

D = (A × n̄) − U

If D is zero or negative the attempts carried the same objects, so the duplicate is a grouping or query problem and resending will not fix it.

If D is positive the archive was asked to keep more copies than the study holds, and the receiver's repeat policy decides what survives.

Redundant bytes ≈ D × B̄, and the display penalty ≈ D × t.

A fictitious case: four attempts of 120 objects, 130 distinct instances, 0.8 MB per object, 40 ms load time.

So D = 480 − 130 = 350, about 280 MB of repeats offered and 14 extra seconds per viewer.

D is an upper bound. A receiver that replaces on equal identity keeps fewer copies. The number picks the bucket; it does not report what the archive did.

The Gate Before Anyone Presses Resend

A retry is cheap to start and expensive to explain. Clear ten items first, each with a named owner.

Attempt count. Sends and retries with timestamps and trigger: automatic, queued job, manual export.

What leaves the box. A queued retry of stored objects, or a fresh export that regenerates them.

Identity sample. One SOP Instance UID per entry, recorded before any further action.

Association result. Accepted, rejected or aborted, with the diagnostic value.

AE Title drift. Any change to calling or called title, address, port or presentation context since the last clean send.

Receiver check. Both titles are mandatory parameters, and the receiver must verify the calling title against its known remote application names; diagnostic 03 means it did not, which is drift rather than storage fault [4].

Queue residue. Jobs still pending from the failed attempt will re-send on their own schedule.

Commitment state. Whether storage commitment was requested and answered, distinct from C-STORE acceptance [5].

Break point. Where the connection ended: association setup, mid-transfer, or after the last object.

Stop rule. Who may call the entry a display artifact, and who signs off consolidation.

Read the Status in Its Own Service

Status 0111H is defined in the message exchange standard as Duplicate SOP Instance, and may carry the affected SOP Instance UID already allocated to another instance [1].

It is absent from the storage response table, which lists 0000, the warnings B000, B006 and B007, and the failure ranges A7xx, A9xx and Cxxx [2].

A duplicate code therefore means nothing without its operation. A duplicate raised on a procedure step message says nothing about which images reached the archive.

Modality Performed Procedure Step reports in progress, discontinued or completed through performed procedure step status (0040,0252) [6]. A discontinued step can look like a second procedure with no extra image stored.

Storage commitment supplies the missing promise. The storage service class obliges the receiver only to accept; commitment adds safekeeping, with duration and retrieve capability documented in the conformance statement [5].

Why the Probe Sits Last

Probe and front-end faults are real, common and costly. They are almost never why a study appears twice, and ranking them high costs a week of wrong work.

A transducer carries probe identification in its EEPROM, which is why a failing chip appears as an unrecognised or misidentified probe on the host. That identifier describes the probe, not the examination.

Study, series and instance identifiers are created by the host and carried in the dataset, and must agree with the message and file-meta copies of the same object [3].

Swapping a probe, reseating a connector or reprogramming an EEPROM cannot create a second study identity. A duplicate entry argues against the probe.

Hardware earns its place on physical evidence: an artifact that survives local review, a probe failing leak or element tests, a connector dropping associations. A duplicate row is not that evidence.

Draw the Responsibility Line Early

Most duplicate disputes are contract questions surfacing as technical ones. Four lines are worth writing while systems are healthy.

Source generation. The modality owner states which export paths exist and which regenerate identifiers.

Ingestion and repeat policy. The archive owner states the effective policy, its version and where it is documented.

Safekeeping. The commitment holder states duration and retrieve capability, per its conformance statement [5].

Clinical completeness. The clinical owner states which objects must be present for the report to stand.

An agreement naming these four owners is cheaper than a dispute settled by guesswork.

The Evidence Pack: Six Files

An incident that closes without a record gets reopened. Keep six items and name the holder of each.

Two identity captures. One object from each entry, showing SOP Class UID and SOP Instance UID.

The stored-object list. What the archive reports for the accession, exported by the PACS administrator.

The association result. Outcome, diagnostic value and the AE Titles actually used.

The queue state. Pending and completed jobs at the moment of the retry, with trigger.

The policy statement. Repeat policy in force at event time, with version, not today's setting.

The decision note. Who concluded what, on which evidence, and what was left open.

Redact patient-linked values before the pack leaves the facility.

Acceptance Checklist

  • The symptom class is written in one sentence: two studies, extra series, extra tiles, or unchanged content.
  • One SOP Instance UID pair is recorded per entry, not an accession number or an image count.
  • Study and series hierarchy is compared separately from the viewer's grouping.
  • The repeat policy in force at event time is on file with its version.
  • An association result exists for each attempt, with AE Titles and diagnostic value.
  • The repeat arithmetic is recorded with the conclusion it supports.
  • A named owner has accepted residual risk wherever evidence is missing.
  • No record was deleted, no identifier rewritten, and no repeat sent as a "fix".

Where GEPROBE Fits

GEPROBE repairs ultrasound probes and supplies parts and systems [7][8]. We do not administer archives, and nothing here is a recommendation about your PACS.

Our published scope is restoration: element-array image quality, articulation, sealing and leak integrity, connector pin repair, EEPROM reprogramming with host recognition checks, lens and membrane, cable and strain relief [7].

The site-stated process is a free 24-hour evaluation with a detailed failure report, a flat-rate quote with repair or replace options, cleanroom restoration, then functional test and imaging verification [7].

That scope matters at the last rung of the ladder, when a physical fault reproduces locally. It is not a route to fixing a duplicate study entry, and no repair changes an archive identifier.

If a probe or board fails its own tests, start at ultrasound probe repair, browse ultrasound parts, or send sanitised symptoms for a quotation.

Sources

  1. DICOM PS3.7 (2026d) C.5.8 Duplicate SOP Instance: status 0111H and affected SOP Instance UID — reviewed 4 November 2026.
  2. DICOM PS3.4 (2026d) B.2.3, Table B.2-1 C-STORE response status values — reviewed 4 November 2026.
  3. DICOM PS3.3 (2026d) C.12.1.1.1 SOP Class UID and SOP Instance UID, agreement with DIMSE and file-meta copies — reviewed 4 November 2026.
  4. DICOM PS3.8 (2026d) Chapter 7, 7.1.1 A-ASSOCIATE parameters: calling and called AE Title, verification, diagnostic 03 — reviewed 4 November 2026.
  5. DICOM PS3.4 (2026d) Annex J.1.1 Storage Commitment scope: no safekeeping promise in the storage service class — reviewed 4 November 2026.
  6. DICOM PS3.4 (2026d) Annex F.1.5 Modality Performed Procedure Step states and status (0040,0252) — reviewed 4 November 2026.
  7. GEPROBE repair services: published scope, evaluation, cleanroom restoration, imaging verification — reviewed 4 November 2026.
  8. GEPROBE ultrasound parts catalogue — reviewed 4 November 2026.