Mammography and Breast Imaging Workflows: Multi-Site Integration, Priors, and AI
A multi-site breast-imaging workflow must deliver current and prior mammography studies, preserve presentation and CAD context, normalize data from different systems, and make results available across PACS, archives, and reading locations. Enterprise integration reduces manual retrieval and format barriers without requiring every facility to operate the same acquisition or PACS platform.
Breast imaging may include full-field digital mammography, digital breast tomosynthesis, ultrasound, MRI, biopsy imaging, reports, and CAD or AI results. These resources often originate in different systems and carry different patient identifiers, procedure codes, metadata conventions, object types, and presentation requirements.
A reliable workflow coordinates orders, patient identity, acquisition, routing, prior retrieval, interpretation, reporting, results distribution, and monitoring. It must also account for the capabilities of each modality, PACS, archive, viewer, reporting system, and AI application.
Why Mammography Requires a Dedicated Enterprise Workflow
Mammography places specific demands on enterprise imaging infrastructure.
Radiologists frequently compare current images with prior examinations to identify interval changes. The American College of Radiology practice parameter for screening and diagnostic mammography identifies comparison with prior breast imaging as an important part of interpretation and recommends obtaining prior studies when needed.
Digital breast tomosynthesis also creates larger multiframe studies than conventional 2D mammography. Those studies can affect transfer times, cache capacity, viewer performance, and the timing of prior retrieval.
Presentation depends on more than pixel data. Laterality, view position, image type, acquisition information, frame relationships, and references to prior or CAD objects can affect how the receiving viewer organizes and displays a study.
Multi-site health systems must therefore coordinate several functions:
- Patient and order reconciliation
- DICOM Modality Worklist delivery
- Current-study routing
- Cross-PACS prior discovery and retrieval
- Metadata normalization
- Legacy format conversion
- Viewer and hanging protocol validation
- CAD and AI integration
- Results distribution
- Exception handling and operational monitoring
The IHE Radiology Technical Framework includes profiles for Mammography Image, Digital Breast Tomosynthesis, Scheduled Workflow, Patient Information Reconciliation, Consistent Presentation of Images, and Import and Display of External Priors. These profiles provide useful standards-based guidance, but each connected product still requires interface and conformance testing.
The End-to-End Multi-Site Mammography Workflow
A production mammography workflow should coordinate the following steps.
- Schedule the examination.
The EHR, RIS, or scheduling system creates the order and communicates the patient, procedure, location, and appointment information to the appropriate systems. - Reconcile patient identity.
The workflow compares identifiers, assigns authorities, and examines demographics and known relationships between enterprise identities. Uncertain matches should enter an exception queue rather than be resolved by an automatic tag change. - Populate the modality worklist.
The modality receives scheduled procedure information through DICOM Modality Worklist or another supported interface. This reduces manual entry and helps maintain consistency between the order and acquired study. - Acquire and validate the current study.
The workflow checks required identifiers, SOP Classes, transfer syntaxes, laterality, view position, image type, and other routing or display attributes. - Search for relevant priors.
The integration layer queries the appropriate PACS, VNA, or external archive based on patient identity, modality, body region, procedure, date range, facility, and radiologist preferences. - Retrieve and prepare priors.
Selected priors are retrieved early enough for interpretation. The workflow may need to normalize metadata or convert legacy representations before sending the studies to the destination system. - Route studies to the reading environment.
Current and prior studies are delivered or made accessible to the breast-imaging PACS, enterprise viewer, workstation, or remote reading location. - Submit eligible studies to CAD or AI.
Routing rules identify the correct studies and send only the images and clinical context required by the approved application. - Validate and distribute results.
Returned CAD or AI results are checked for patient and study linkage, expected object type, completeness, and destination compatibility before delivery. - Monitor the workflow.
Operations teams track missing priors, failed matches, conversion errors, delayed transfers, rejected objects, unavailable destinations, and undelivered results.
How Relevant Priors Are Retrieved Across Sites
Relevant priors should be available in the reading environment before the radiologist begins interpretation. Portable media may still serve as a fallback, but it should not remain the primary enterprise strategy.
A cross-site prior workflow typically follows this sequence:
- An order, assignment, or incoming study triggers a prior search.
- The workflow resolves the patient across local and external identity domains.
- It queries connected PACS and archives for candidate studies.
- Rules select priors based on modality, body region, procedure, date range, facility, and reading requirements.
- The workflow retrieves the selected studies from one or more sources.
- It validates identifiers, object types, transfer syntaxes, and required display metadata.
- It converts or normalizes data when the destination requires it.
- It stages the studies in the receiving PACS or makes them available through an enterprise access layer.
- It records success, partial completion, or failure and directs exceptions to the appropriate team.
A patient name or medical record number match alone is not enough. Enterprise workflows should use assigning authorities, demographic attributes, enterprise identifiers, and governed match rules. A questionable match should require review.
Prefetch Relevant Priors Process
What Are BTO, CTO, and SCO?
Not every tomosynthesis study created by a legacy or vendor-specific workflow is represented in the format a receiving breast viewer expects.
| Term | What it represents | Interoperability concern | Recommended workflow |
|---|---|---|---|
| BTO | The DICOM Breast Tomosynthesis Image Storage SOP Class designed for reconstructed DBT image volumes | The receiving PACS and viewer must support the object, transfer syntax, frames, dimensions, and breast-specific attributes | Preserve and validate the BTO object when it is already compatible |
| CTO | A CT Image Storage object used by some legacy or vendor-specific workflows to represent tomosynthesis content | CT Image Storage is a standard DICOM SOP Class, but a breast viewer may not interpret vendor-specific tomosynthesis structure or presentation context correctly | Characterize the source representation and convert it only when the destination requires a validated BTO object |
| SCO | A Secondary Capture object containing a rendered image representation | Secondary Capture may not preserve the acquisition structure, breast-specific metadata, frame relationships, or processing information required for full DBT functionality | Determine whether the rendered representation is clinically sufficient or whether a validated BTO conversion is required |
The DICOM Standard defines CT Image Storage, Secondary Capture, Digital Mammography, Breast Tomosynthesis, and other SOP Classes. The interoperability problem is not simply whether an object is DICOM. The issue is whether the object preserves the structure and context that the destination system needs.
A conversion workflow should document:
- Source SOP Class and transfer syntax
- Private tags and vendor-specific attributes
- Frame order and dimensional relationships
- Patient, study, series, and instance identifiers
- Laterality and view position
- Image type and processing state
- Source attribution and provenance
- Destination PACS and viewer requirements
- Image fidelity validation
- Exception and rollback procedures
No organization should assume that every CTO or SCO object can be converted automatically. The converted output must be tested with each receiving PACS and diagnostic viewer.
SCO/CTO to BTO Conversion mammography workflow with Dicom Systems
Presentation and Hanging Protocol Dependencies
| Dependency | Why it matters | What to validate |
|---|---|---|
| Patient and accession identifiers | Current and prior studies must attach to the correct patient and order | Identity domains, assigning authorities, accession rules, merges, and overlays |
| SOP Class and transfer syntax | PACS and viewers may support different image representations | Storage, retrieval, decompression, display, and export behavior |
| Laterality and view position | Hanging protocols use these values to organize breast images | Left and right breast, CC, MLO, and additional diagnostic views |
| Image type and processing state | Viewers may treat for-presentation and for-processing images differently | Expected display behavior and processing restrictions |
| Series and frame relationships | DBT viewers rely on multiframe structure and dimensional context | Frame order, reconstructed slices, projection images, and linked series |
| Priors and facility identifiers | Hanging rules may distinguish current studies from priors or group studies by location | Date logic, facility codes, procedure codes, and prior status |
| CAD and AI references | Results must remain associated with the correct source images | Referenced studies, series, instances, coordinates, and result visibility |
| Viewer configuration | Each viewer implements layouts and supported objects differently | Vendor conformance statements, hanging protocols, diagnostic display, and user acceptance testing |
Health systems should test representative studies from every modality model, software version, location, and legacy archive involved in the workflow.
Routing Mammography Studies to CAD and AI
A production CAD or AI workflow requires more than sending images to an algorithm.
The workflow should:
- Identify eligible studies using approved routing criteria.
- Confirm that the study contains the required image types and clinical context.
- Route the study to the correct on-premises, cloud, or hybrid application.
- Track submission status, processing time, and failures.
- Validate the returned patient, study, series, and instance references.
- Confirm that the result uses a format the destination supports.
- Deliver the result to the appropriate PACS, viewer, reporting system, or EHR.
- Preserve the application, model, version, processing time, and other required provenance.
- Retry recoverable failures and route unresolved exceptions for review.
The DICOM Standard defines a Mammography CAD Structured Report, but not every application produces it, and not every receiving system displays it. Organizations must validate both ends of the interface.
| Result representation | Typical use | Validation requirement |
|---|---|---|
| Mammography CAD SR | Structured findings, coordinates, measurements, and image references | Confirm that the receiving PACS and viewer support the object and its referenced images |
| Other supported DICOM SR | Structured measurements or observations | Validate templates, codes, references, and display behavior |
| Presentation or annotation object | Marks, overlays, or display instructions | Confirm coordinate systems, source-image references, and viewer support |
| Secondary Capture | Rendered result image when structured consumption is unavailable | Confirm clinical usefulness and avoid treating the rendered image as the complete structured result |
| HL7, FHIR, or vendor API output | Status, findings, measurements, links, or workflow events | Map the result to the correct patient, order, report, and destination workflow |
A result should not be considered delivered merely because the destination accepted a network transaction. The workflow should verify that clinicians can find and use the result in the intended application.
Multi-Site and Cross-PACS Access Patterns
Health systems do not need to standardize every facility’s PACS to improve mammography access. They can use several integration patterns.
| Pattern | How it works | Best fit |
|---|---|---|
| Centralized archive | Sites send studies to a central PACS or VNA | Organizations consolidating storage and reading |
| Federated query and retrieve | An integration layer queries multiple PACS and consolidates available results | Organizations retaining regional or specialty PACS |
| Staged prior delivery | Selected priors are retrieved and sent to the reading PACS before interpretation | Workflows that require local diagnostic-viewer access |
| Enterprise viewer access | Authorized users open studies from connected archives through a shared viewing layer | Broad clinical access and on-demand review |
| Hybrid model | Current studies, recent priors, and long-term history use different access patterns | Large organizations with varied performance and retention requirements |
The architecture should account for patient identity, network performance, transfer timing, duplicate studies, source attribution, viewer compatibility, and failure handling.
Operational Monitoring and Workflow KPIs
A breast-imaging workflow needs continuous monitoring. A successful test during implementation does not prove that the workflow will remain reliable after upgrades, site additions, routing-rule changes, or growing study volumes.
Useful operational measures include:
- Percentage of scheduled examinations with required priors available before interpretation
- Prior searches that return no candidates
- Partial or failed prior retrievals
- Patient-match exceptions
- Incorrect or missing laterality and view-position values
- Rejected DICOM objects
- CTO or SCO conversion success and failure rates
- Transfer latency by site and destination
- Time from acquisition to availability in the reading environment
- AI or CAD submission and result-delivery latency
- Results accepted by a destination but unavailable in the clinical viewer
- Duplicate study and duplicate result rates
- Queue depth and oldest pending transaction
- Availability of each connected PACS, archive, viewer, and AI application
Dashboards should link metrics to actionable exceptions. Each exception needs an owner, severity, status, audit history, and defined resolution path.
MQSA and Information-Workflow Governance
The FDA’s amended Mammography Quality Standards Act regulations, including federal breast density notification requirements, took effect on September 10, 2024.
The amended requirements address mammography reports, patient lay summaries, breast density information, communication timing, recordkeeping, and the transfer of mammography records. For example, the FDA requires accelerated communication when the final assessment is suspicious or highly suggestive of malignancy and sets requirements for examinations that remain incomplete because prior mammograms are needed.
An enterprise workflow can support these responsibilities by maintaining accurate patient and examination context, preserving reports and images, tracking delivery, supporting retrieval, and recording operational evidence. Technology does not establish compliance by itself. Each organization must define its policies with clinical, compliance, legal, and information-security stakeholders.
Mammography Workflow Implementation Checklist
Before deployment, document and test:
- Every mammography, DBT, ultrasound, MRI, and biopsy acquisition system
- PACS, VNA, EHR, RIS, viewers, reporting systems, and AI or CAD applications
- DICOM and non-DICOM interfaces
- Supported SOP Classes and transfer syntaxes
- Patient-identity domains and assigning authorities
- Order, accession, merge, and correction workflows
- Relevant-prior selection and retrieval rules
- BTO, CTO, SCO, and vendor-specific representations
- Approved metadata transformations
- Hanging protocol and viewer dependencies
- AI and CAD result formats
- Routing priorities and network constraints
- Exception ownership and escalation
- Monitoring thresholds and service-level targets
- Upgrade, regression, failover, and rollback tests
- Retention, record-transfer, privacy, security, and audit requirements
Frequently Asked Questions
How is mammography integrated with enterprise imaging?
Mammography integration connects acquisition systems, PACS, VNA, EHR, RIS, viewers, reporting systems, and AI or CAD applications via standards-based, vendor-supported interfaces. The integration layer coordinates identity, worklists, routing, priors, normalization, result delivery, and monitoring.
How are priors accessed across sites?
A query and retrieval workflow searches connected PACS and archives, reconciles the patient across identity domains, selects relevant studies, and retrieves or exposes those studies to the reading environment. The workflow should track partial results, failed retrievals, and questionable patient matches.
What are BTO, CTO, and SCO formats?
BTO is the DICOM Breast Tomosynthesis Image Storage SOP Class designed for reconstructed DBT volumes. CTO commonly refers to CT Image Storage used by some legacy or vendor workflows, while SCO refers to Secondary Capture. CTO and SCO can be valid DICOM objects, but they may not preserve the breast-specific structure or presentation context another viewer requires.
Can breast-imaging AI results be routed across PACS?
Yes, when the source application, integration layer, destination PACS, and viewer support compatible interfaces and result formats. The workflow must preserve patient and study linkage, validate references, route the result to the correct destination, and monitor whether the result becomes clinically accessible.
How are multi-site mammography workflows consolidated?
Organizations can centralize studies, federate searches across PACS, stage priors in a reading system, provide enterprise-viewer access, or combine these approaches. A shared integration layer coordinates the systems without requiring every facility to replace its PACS or acquisition equipment.
How Dicom Systems Supports Multi-Site Mammography Workflows
Dicom Systems supports breast-imaging workflows through the Unifier Enterprise Imaging Platform. Health systems choose which mammography modalities, PACS, VNA, EHR, RIS, viewers, reporting systems, archives, and AI applications to connect. Dicom Systems configures the workflow around those systems.
The Dicom Systems mammography solution supports 2D mammography, 3D tomosynthesis, ultrasound, MRI, and related enterprise imaging workflows.
Relevant prior retrieval
Unifier can query connected PACS and archives for relevant priors, consolidate available studies, apply selection rules, and send the required studies to the reading environment. Organizations can define criteria based on modality, body region, date range, location, and other available attributes.
Patient matching
Patient-matching workflows help reconcile identifiers across connected systems. Dicom Systems can configure match rules, identity transformations, and exception handling based on the organization’s identity domains and policies.
Modality worklist and order integration
Unifier supports DICOM Modality Worklist and healthcare messaging workflows that connect scheduling, acquisition, and downstream imaging systems. These connections reduce manual entry and help maintain the patient, order, and procedure context.
SCO and CTO to BTO conversion
Dicom Systems provides SCO and CTO-to-BTO conversion for validated workflows in which legacy or vendor-specific representations are incompatible with the destination breast-imaging system. The conversion can be combined with routing, prior retrieval, and metadata transformation.
Routing and performance
Unifier can route current studies and priors across selected destinations, prioritize traffic, apply supported compression, and use load-balancing workflows to support high-volume environments. Configuration depends on the source objects, network, destination capabilities, and clinical timing requirements.
CAD and AI integration
Dicom Systems can connect breast imaging workflows to CAD and AI applications via AI Conductor and other Unifier capabilities. The workflow can route eligible studies, manage supported result interfaces, deliver outputs to selected clinical systems, and monitor failures and exceptions.
Monitoring and exception management
Dicom Systems configures operational monitoring around the connected systems and workflows. Teams can track routing failures, prior-retrieval issues, patient-match exceptions, conversion errors, unavailable destinations, and undelivered results.
Meet with a Dicom Systems workflow expert to review the mammography modalities, PACS, VNA, EHR, viewers, reporting systems, archives, and AI applications you need to connect. Define the prior-retrieval, patient-matching, BTO conversion, routing, presentation, result-delivery, exception, and monitoring requirements for your environment.