AI Workflow Solution

AI Workflow Solution

Introduction

Today, we are going to be looking at the Lorebridge AI workflow using our Compass Router. We're going to utilize Compass to facilitate two critical tasks:

1. De-identifying images as they are routed from a site or hospital to an AI algorithm server, which might be operating off-site or in the cloud.
2. Re-identifying the reported image from the algorithm server back to the site Compass, reintegrating the original patient demographics, and finally sending it to the local PACS (Picture Archiving and Communication System).

The aim is to ensure that the original patient information never leaves the site's hospital premises. It is encrypted at the hospital and can only be decrypted at the hospital.

Steps Involved

1. De-Identification Process

   • The Study Instance UID is used as the unique key for this demonstration. This is the only field sent in the clear to the AI server and back to the hospital. Other keys could be chosen depending on the setup.

2. Compass Router Interface

   • The main user interface (UI) of the Compass Router is the window in the center. Additional windows visible are parts of the Lorebridge suite called PowerTools, used to demonstrate sending images from a simulated modality to the AI algorithm server and back to the local PACS via Compass.

3. Sending Images

   • Images are simulated and sent from a modality. For instance, patient John Doe with MRN PID1234. These images are routed to the AI workstation through Compass.
   • At the AI server, the images are de-identified so no patient information is visible off-site.

4. Re-Identification Process

   • Post processing, an image report is generated by the AI algorithm server and sent back to the site Compass for re-identification.
   • The original patient data remains encrypted and is only re-utilized at the site hospital.

5. Validation

   • The system checks if the re-identified report matches the original patient demographics before sending it to the local PACS.
   • In the local PACS, the report is placed into the correct patient jacket.

Demonstration Example

Simulation:

• Sending MRI images of John Doe (PID1234) to Compass at the hospital site.
• Image routing de-identified to the AI workstation.

AI Server:

• De-identified patient data visible.
• AI generates a DICOM secondary capture image report using the Study Instance UID as a unique key without any patient information.

Compass Re-Identification:

• AI server sends the de-identified report back to site Compass.
• Report re-identified with the original patient demographics and forwarded to the local PACS.
• Local PACS confirms demographics match, placing report in the correct patient jacket.

Keywords

• Lorebridge AI Workflow
• Compass Router
• De-identification
• Re-identification
• Image Routing
• AI Algorithm Server
• Study Instance UID
• Local PACS
• Patient Safety
• Data Encryption

FAQ

Q1: What is the purpose of the Lorebridge AI workflow with Compass Router?

A1: The primary goal is to securely de-identify images before sending them off-site to an AI algorithm server and then re-identifying them with original patient demographics upon return, ensuring patient information security.

Q2: What unique key is used in this demonstration for tracking the images?

A2: The Study Instance UID is used as the unique key in this demonstration.

Q3: Can another unique identifier other than the Study Instance UID be used?

A3: Yes, other unique identifiers can be chosen based on the requirements and setup.

Q4: How does the system ensure the original patient information doesn't leave the hospital site?

A4: The patient information is encrypted at the hospital site and only decrypted and re-identified at the same site.

Q5: What are the components of the Lorebridge suite used in this demonstration?

A5: The demonstration uses the Compass Router as the main UI and other parts of the Lorebridge suite called PowerTools for image handling and routing.

Q6: How is patient information managed after AI processing?

A6: Post-AI processing, the report is generated without patient details and sent back to the site Compass, where it gets re-identified with original patient data and sent to local PACS for accurate patient record keeping.

Q7: What ensures that the re-identified report matches the original patient data?

A7: The system validates the demographics before sending the report to the local PACS, ensuring the report is placed in the correct patient jacket.