Standards in Digital Pathology: Status quo

DOI: https://doi.org/10.47184/tp.2023.01.01

Recent advancements in whole slide digital imaging technologies brought and will bring significantly altered traditional workflows within pathology laboratories. A diagnostic digital pathology requires the deep integration of pathology laboratory and imaging workflows, including order entry, result reporting, image storing, image manipulation, and image management. As far as different systems from different vendors are required for those integrated workflows, a common understanding for the workflow between the different partners is inevitable. Standards are the formalized representation of that common understanding.The status quo of four main classes of international reference standards for their use in digital pathology is described.

Keywords: Digital Pathology, Workflow, HL7 Standard family, IHE, DICOM, LOINC, SNOMED-CT

Introduction

Worldwide implementations of Digital Pathology solutions have evolved and advanced rapidly, and data support the benefits of digital pathology over routine microscopy [1, 2], especially the downstream analytical benefits of a digital format [3]. FDA approval of the first WSI system for primary diagnosis [4] has reignited focus on use cases, and value propositions for various vended solutions that in aggregate would support digital pathology. However, currently, there is minimal formal literature that speaks to optimal interactions among vended solutions [5].

Anatomic pathology laboratory information systems (AP‑LIS) focus on the production, storage, and conveyance of the diagnostic interpretation of stained tissue. AP‑LIS is transitioning to become modules of larger Electronic Health Record systems, and during this transitional period, the implementation of these systems is quite variable across the market. WSI will drive additional evolution and/or reinvention of the AP‑LIS. There is more to digital pathology than supporting WSI and associated instrumentation, but this innovation will be a catalyst for change and an enabler of significant evolution of workflow and advancement of the field [5].

As far as different system components are required to establish a workflow that incorporates capabilities for image acquisition, storing and displaying, as well as for image analysis, second opinions and result reporting, a common understanding of those processes between all partners involved is inevitable for getting interoperability. Standards are the formalized representation of that common understanding.

 

Classes of standards

For Digital Pathology, a wide variety of national and international “standards“ are currently in use. Only a very small minority of them fulfill the needs described above, namely to guarantee vendor-neutral interoperability between subsystems of pathology, imaging and clinics.

Four classes of reference standards are required:

  • Communication standards for workflow and reporting
  • Standards for imaging and images
  • Document standards
  • Standard terminologies.

There are only three families of reference standards, used internationally: the HL7 standard family, DICOM, and PDF, added by three reference terminologies: LOINC, UCUM and SNOMED-CT. Additionally, a few international classification systems, like ICD-10, ICD-11, and ICD-O can also be assumed as reference terminologies.

These standards are very comprehensive, covering the entire field of medicine, not suited to serve as implementable software solutions directly. Instead, they have to be constrained for specific use cases. This is done by Standard Developing Organizations (SDO) like HL7, DICOM, LOINC or SNOMED-CT themselves or by the IHE in form of Implementation Guides or Integration Profiles. The SDOs and IHE publish each profile through a well‑defined process of public review and trial implementation and gathers profiles that have reached final text status into an HL7 Implementation Guide or an IHE Technical Frameworks.

The HL7 standard family

HL7 is the most used standard in hospital and laboratory information systems. It comprises the message standard HL7 version 2.x, the document standard HL7 version 3 / CDA (Clinical Document Architecture), and the very modern standard FHIR. Only in the last few years it starts to replace the widely used xDT standard in the out-patient sector in Germany. FHIR is recently becoming the more or less “official“ standard in Germany by legislation and the actual activities of major players in the field, without replacing the v2.x at larger extent. Whereas HL7, v.2.x and CDA are message or document-based and event-driven, FHIR is data-based and query-driven. Most of the IHE Integration Profiles of the PaLM domain (Pathology and Laboratory Medicine) focus on HL7 version 2.x and CDA. The Medical Information Objects (MIO) and the Medical Informatics Initiative (MII) in Germany use FHIR resources. All HL7 family members can be made interoperable with each other. Due to its high complexity in implementation, CDA has not been used in Germany at a wider scale so far.

DICOM

DICOM has been developed since the seventies of the 20th century for radiology, but it is applicable for imaging processes and images in all medical disciplines. It is still very uncommon in pathology. Especially the DICOM Supplements 122 for specimen workflow, 145 for whole slide imaging, and 222 for bulk annotations, are specifically dedicated for Digital Pathology. As to avoid parallel standard development, IHE and DICOM are closely cooperating between the PaLM and the Workgroup 26 for years. The DICOM SR (Structured Reporting) can be used for Digital Pathology, too, but this is recently not used.

The PDF standard family

PDF is the most important document standard, in contrast to CDA widely used in hospital and pathology information systems. As PDF/A-3 or A-4 it is especially suited to carry text and formatting data together with embedded html or xml (e. g. the FHIR instances), in such a way it is currently the optimal document standard for structured reporting.

Reference terminologies

When real interoperability is desired, the use of reference terminologies is inevitable. Internationally, LOINC, UCUM, and SNOMED-CT are the most important ones. Whereas UCUM is used for coding the units of measurement values, the other both are nearly completely covering the entire field of laboratory and clinical medicine. Especially, SNOMED-CT is a highly complex terminology with ontological elements, so that their codes can be interpreted in clinical contexts by machines.

Profiling standards for digital pathology

It is the profiling only, what makes a standard to a sufficient condition for interoperability. Especially the IHE is focused to the definition of use-case based profiles, which combine and constrain base standards like HL7, DICOM and reference terminologies for the very specific field of application. But also the HL7 base standard is use-case specifically constrained and added by terminology bindings by national and international HL7 organizations in HL7-Implementation Guides.

Currently, the following profiles for digital pathology use cases are of interest (either in final status (green), in trial use (yellow), or in development (orange)), as shown in table 1 [6, 7, 8, 9].

Table 1: Current profiles for digital pathology use cases.

Conclusions

For most of Digital Pathology use cases, international reference standards and terminologies have been profiled during the last few years. Thus, vendors of Digital Pathology system components and their customers could rely on interoperable systems if the appropriate system components had been tested successfully in so-called connectathons. It should not depend on whether a pathology institution is relying on a vendor that offers an all-in-one product or on a mixture of system components exactly meeting the vendors demands. Digital Pathology is very dynamic; therefore the status quo of their standards will be dynamic, too.

Author
Prof. Dr. Gunter Haroske
Federal Association of German Pathologists Digital Pathology Commission, Berlin