How do I set up a DICOM workflow tool for lesion segmentation and tumor measurement in our PACS environment?

Setting up a DICOM workflow for lesion segmentation and tumor measurement in a PACS environment usually means combining image ingestion, segmentation/annotation, measurement/reporting, and storage/output back to PACS in a standards-based way.

1) Define the clinical workflow first

Decide what the system must support:

• Modalities: CT, MR, PET/CT, US?
• Use case: lesion segmentation, RECIST measurement, volumetrics, follow-up comparison?
• Users: radiologists only, oncology team, research users?
• Output needed:
  • DICOM SEG objects
  • DICOM SR / TID 1500 measurement reports
  • screenshots/PDF
  • structured export to LIS/RIS/EHR or data warehouse

2) Make sure PACS interoperability is standards-based

Your workflow tool should ideally support:

• DICOM C-STORE: send/store images and results
• DICOM C-FIND / C-MOVE or C-GET: retrieve studies
• DICOMweb if available:
  • QIDO-RS for query
  • WADO-RS for retrieve
  • STOW-RS for storing results
• DICOM SEG for segmentations
• DICOM SR for measurements and structured reporting
• Ideally SOP class support for the specific modalities and derived objects you need

3) Choose the workflow architecture

Common patterns:

A. PACS-integrated workstation

• A radiology workstation runs the segmentation software
• Reads studies from PACS
• Saves results back to PACS
• Best for small/medium deployments

B. Server-based processing pipeline

• PACS sends studies to a processing server
• Server performs AI segmentation and/or manual review
• Results are pushed back to PACS
• Better for scale, automation, and multi-user access

C. Hybrid

• AI preprocessing on server
• Radiologist review/approval on workstation
• Approved results archived back to PACS

4) Core components you’ll need

Typical stack:

• PACS / VNA
• DICOM router or integration engine
  e.g., Orthanc, dcm4chee, Mirth Connect, vendor gateway
• Segmentation/measurement application
  • supports manual, semi-auto, or AI-driven contouring
• Database
  • stores study metadata, measurements, task status
• Viewer
  • shows original images plus contours and measurements
• Reporting/export layer
  • sends structured outputs back to PACS/EHR

5) Pick tools that support DICOM SEG and SR

When evaluating software, check that it can:

• Import multi-slice CT/MR studies
• Allow contour editing
• Generate DICOM SEG
• Generate DICOM SR for lesion measurements
• Keep references to source images and frame numbers
• Support lesion tracking over time
• Store results as derived objects in PACS

If the software only exports screenshots or proprietary files, that’s usually a limitation for clinical PACS workflows.

6) Configure identity and matching carefully

Measurement workflows depend on correct study/patient matching:

• Use Accession Number, Study Instance UID, and Patient ID
• Ensure consistent demographics from HIS/RIS
• Avoid duplicate patient creation
• Validate study timepoints for follow-up comparisons
• Use hanging protocols or lesion tracking by series/study UID

7) Plan the lesion measurement approach

For tumor measurement, define whether you need:

• Linear measurements: longest diameter, RECIST
• Bidimensional measurements
• Volumetric segmentation
• SUV measurements for PET
• Whole-lesion burden or organ-specific quantification

For oncology, many workflows support:

• Baseline lesion selection
• Target lesion measurements
• Follow-up propagation
• Response assessment categories

8) Integrate with reporting and results storage

Good practice is to store:

• Segmentation masks as DICOM SEG
• Quantitative results as DICOM SR
• Optional human-readable summary in the report system
• Links back to the source study and series

Also decide whether results should appear in:

• PACS viewer
• Radiology report
• Oncology dashboard
• Research export

9) Security, compliance, and audit

Because this is clinical imaging data, you’ll need:

• Authentication and role-based access control
• TLS for DICOM/DICOMweb where possible
• Audit logs for segmentation changes and approvals
• PHI protection and retention policies
• If applicable, HIPAA/GDPR compliance review
• Vendor validation if used for clinical decision-making

10) Test with a pilot workflow

Before production:

1. Import a small set of test CT/MR studies
2. Run segmentation and measurements
3. Verify the objects are valid DICOM SEG/SR
4. Confirm they appear correctly in PACS
5. Validate follow-up study comparison
6. Get radiologist sign-off on usability and measurement accuracy

11) Operational checklist

A practical deployment checklist:

• PACS supports DICOM storage of derived objects
• Segmentation tool supports DICOM SEG/SR
• C-FIND/C-MOVE or DICOMweb retrieval works
• Patient/study matching is reliable
• Security and audit logging are enabled
• QA process for measurement accuracy is defined
• Radiologist review/sign-off workflow is in place
• Results are visible in PACS and/or report system

12) Common software/platforms to evaluate

Depending on your environment, people often look at:

• Orthanc or dcm4chee for DICOM routing/archive
• 3D Slicer for segmentation workflows
• Vendor PACS measurement modules
• AI segmentation platforms with DICOM SEG export
• Mirth Connect for integration messaging
• DICOMweb-enabled viewer/workflow platforms

---

Recommended implementation path

If you want a reliable clinical setup:

1. Start with a small pilot
2. Use DICOMweb or standard DICOM routing
3. Ensure outputs are DICOM SEG + SR
4. Add manual radiologist review
5. Validate against your PACS and reporting workflow
6. Scale only after QA and compliance approval

If you want, I can also help you with one of these next:

• a reference architecture diagram
• a vendor-neutral system design
• a step-by-step setup using Orthanc + 3D Slicer
• a checklist for PACS/DICOM integration testing