Facial Recognition – Not just the surface

Use Case

  1. When a patient is admitted, a front-facing photo is taken.
  2. The photo is stored either locally or in Azure Blob Storage.
  3. The URI for the image is inserted into an HL7 Version 2.x OBX-5 message.
  4. Once the Processor Service receives the message it is added to the HL7 FHIR Patient Resource.
  5. When the patient is brought into the Operating Room, a facial scan is performed.
  6. It is verified using the Azure Cogitative Face API Service.

Solving the problem

I have been developing an extension device which can be attached to a Cell Phone or Pad.

I have several 3-D Cameras and am also working with the Azure Kinect SDK.
I am currently training the device software to detect not just the face, but the complete muscle and bone structure. The following video shows what I am targeting.

Patient Identification Device

  • The device will have it own camera or can be used with the phone or pad own camera.
  • It will use the phone or pads WiFI to connect to a IoT Edge Server
  • The server could connect to Azure to synchronize the Patient images.
  • The server would also store the patients image .
  • Since the server has only one purpose, a Raspberry PI 4 or a Nano Jetson with the Extension Board for a 1-TB SSD would be all that is needed.
  • The Server would run Linux and be configured as an HL7 FHIR Server and API.
  • It would only contain the FHIR Resources for Medical Devices and Patient data.
  • All data would be encrypted in flight and at REST.

I am not in the business of selling products. I normally open-source everything.
I plan on providing a complete tutorial for those that want to build the recognition software. I don’t want to see it ending up being sold by a device manufacturer.

Last modified: March 16, 2020



Leave a Reply