PACS migration project guide

PACS Migration Checklist: Planning, Transfer, Validation and Sign-Off

Use this checklist to turn a large medical imaging transfer into a controlled programme with defined scope, observable execution, managed exceptions and defensible acceptance evidence.

Reviewed and updated 29 July 2026

1. Define scope and acceptance criteria

Start with a written statement of which patients, studies, modalities, facilities and date ranges belong in the migration. Define exclusions and decide how prior studies arriving during the project will be handled. A transfer queue reaching zero is not an acceptance criterion; the project needs measurable evidence that the agreed scope was processed.

Name the authoritative source systems and destination archive.
Document included and excluded study populations.
Agree how counts, identifiers, exceptions and clinical access will be validated.
Assign owners for source PACS, destination PACS, network, application and project decisions.

2. Build a source study inventory

Create a durable study-level inventory before large-scale transfer begins. Record identifiers that can be compared later, such as Patient ID, Study Instance UID, accession number, study date, modality and image count where available. The inventory becomes the control ledger for planning, execution and reconciliation.

Retain the discovery date, source AE Title and query criteria.
Identify duplicate Study Instance UIDs and incomplete demographics.
Separate inaccessible, corrupt or unsupported objects for review.
Baseline total patients, studies, series and instances where the systems expose reliable counts.

3. Validate DICOM and network readiness

Test the full route between the migration controller, source archive and destination. Confirm AE Titles, ports, firewall rules, routing and destination storage behaviour. A successful C-ECHO proves basic association only; representative C-FIND, C-MOVE and C-STORE tests are still required.

Confirm calling and called AE Titles at every endpoint.
Measure practical throughput during representative clinical load.
Verify timeouts, concurrent association limits and retry behaviour.
Ensure clocks and log timestamps are synchronized for later investigation.

4. Run a representative pilot

Choose a pilot set that reflects the archive, not merely its easiest studies. Include multiple modalities, old and recent studies, small and large series, compressed objects and known edge cases. Use the pilot to refine controls and acceptance evidence before scaling up.

Record the exact pilot inventory and expected destination outcome.
Measure throughput without disrupting clinical operations.
Validate image availability through the destination viewer or clinical workflow.
Review failed, partial and duplicate outcomes before approval to scale.

5. Execute in controlled batches

Divide the migration into observable work units by date, facility, modality or another stable rule. Controlled batches make it easier to pause, investigate and resume without losing the relationship between planned scope and actual outcomes.

Define schedules and rate limits with operational stakeholders.
Track queued, running, completed, failed and skipped studies separately.
Keep original identifiers and immutable event timestamps.
Do not treat a successful move request alone as proof of complete receipt.

6. Manage exceptions explicitly

Every exception needs a visible state, reason and next action. Repeated automatic retries can hide configuration faults and create load. Classify failures first, then apply retry, remediation, approved exclusion or escalation as appropriate.

Distinguish network, association, source-read, object and destination-rejection failures.
Retain the attempt count and latest technical response.
Record manual stop, skip and retry decisions with user and time.
Prevent unresolved exceptions from disappearing into a general completed total.

7. Reconcile and validate

Compare the original inventory with destination evidence at study level. Count comparisons are useful alarms, but identifiers are stronger: identical totals can conceal both missing and unexpected studies. Confirm clinical usability for a representative sample and all remediated exceptions.

Match Study Instance UID wherever possible.
Account for every expected study as received, approved exception or unresolved exception.
Investigate unexpected destination studies separately.
Retain query outputs, reports and sampling evidence.

8. Produce acceptance and retirement evidence

Create a final package that a future reviewer can understand without reconstructing the project from application logs. Only then should the organisation decide whether the legacy archive can be made read-only, retained for a defined period or retired.

Scope and acceptance criteria with approvals.
Source inventory and destination reconciliation results.
Exception register with final disposition.
Pilot and production summaries, operational logs and sign-off record.
Legacy access, retention, rollback and decommissioning plan.