Digital Pathology integration

We can identify two basic phases in a Digital Pathology reference workflow:

1. Whole Slide Image (WSI) file and metadata ingestion into the archive

2. Open VIEW on a specific digital slide (or case) with an in-context call

Data ingestion

Actors:

• Digital slide producer (e.g. the WSI scanner)

• Metadata owner system (e.g. Laboratory Information System)

• FEED

• PACS

Note: we assume that there is an integration between the digital slide producer and the metadata owner system.

1. The digital slide producer writes the WSI file into a shared folder (e.g. NFS folder);

2. FEED sees that a new WSI file has been written in the folder;

3. FEED queries the metadata owner system for patient and case information;

4. FEED sends the WSI file and the metadata to the PACS component, where the digital slide and metadata are stored and indexed.

Metadata retrieval transaction

With this transaction, the FEED script will retrieve the metadata from the metadata owner system.

The request type will be an HTTP GET and will contain only one parameter, the slide barcode identifier, that can be derived from the name of the WSI file:

/service/path/metadata?slide=XXXX

In the GET request, it is possible to include a header with a pre-shared static API key as an authorization mechanism (e.g. Authorization: API-KEY).

The service should return the following data in JSON format:

LIS response body example:

{
  PatientID: "PID34125",
  PatientName: "TURNER^KATIE",
  PatientBirthDate: "19750902",
  PatientSex: "F",
  AccessionNumber: "24-H-00123",
  SpecimenIdentifier: "S920939933800092655716259",
  SpecimenAlias: "A",
  SpecimenProcedure: "Breast Biopsy",
  SpecimenBodySite: "BREAST",
  BlockIdentifier: "S906723612258515086899",
  BlockAlias: "A-1",
  BlockProcedure: "Margin"
  SlideIdentifier: "S899706197241433574521",
  SlideAlias: "A-1-A",
  SlideStainCode: "H&E"
}

Virtual tray example and details:

WSI image availability callback

The PACS component can be configured to send an PACS webhooks | Image-Availability-Message to a third-party system to notify that the slides have been correctly imported into the system and are ready to be viewed.

For Digital Pathology, the image availability message will contain one series item for each digital slide, and the slide barcode identifier will be placed in the ContainerIdentifier.

The system will send a message after every new slide the archive receives, and the message will contain information about all the slides in the case.

VIEW integration

An external system like the LIS can open VIEW on a specific digital slide using the VIEW APIs | Access-Token-API-(GET).

In a typical use case, VIEW can be opened on a digital slide by specifying the accNum and containerIdentifier query parameters.

Additional integrations

VIEW allows additional integrations with third-party systems like the LIS.

Webhook events

VIEW notifies a third-party system every time a digital slide status changes. Details can be found in the VIEW webhooks section.

AI algorithms

It is possible to integrate third-party AI algorithms to receive and display findings on processed digital slides. This type of integration necessitates a project-specific analysis.

Snapshots

VIEW can send a snapshot of a digital slide taken by the user to a third-party system. 

The viewer will forward a POST request to the service configured in the NgvConfiguration table under the name ApSnapshotUrl. If the field is empty, no message will be forwarded.

The request will contain the header Content-Type: application/json while its body will look like the following:

{
    "username": "Administrator",
    "timestamp": "2023-09-04T09:56:49.6213677",
    "wholeslide":
    {
        "studyInstanceUID": "1.2.826.0.1.3680043.2.619.161.1027041362",
        "seriesInstanceUID": "1.2.826.0.1.3680043.2.619.1612.1688546606484.1",
        "accessionNumber": "I20-230705",
        "containerIdentifier": "I20-230705-A-1-2"
    },
    "content":
    {
        "contentType": "image/jpeg",
        "caption": "this is a caption",
        "data": <base64 encoded image>,
        "magnification": 5.434888974179961,
        "region":
        [
            {
                "x": 64332.641290268315,
                "y": 6379.44867537442
            },
            {
                "x": 71111.0686279783,
                "y": 6379.44867537442
            },
            {
                "x": 71111.0686279783,
                "y": 8808.201141654868
            },
            {
                "x": 64332.641290268315,
                "y": 8808.201141654868
            },
            {
                "x": 64332.641290268315,
                "y": 6379.44867537442
            }
        ]
    }
}

Focusing on the object content, it will contain:

• the contentType of the snapshot, always set to image/jpeg;

• the caption inserted by the user;

• the image exported in Base64 in the string data;

• the zoom level (magnification) at the time the image was exported;

• the coordinates (region) relative to the exported image.

The remaining shared information is related to the session (username and timestamp) and case or slide identifiers.