Minutes Data Working Group 26 Oct 2020

Agenda

• Review feedback from MONAI Bootcamp
• Review Wenqi's finding on Ivadomed, currently using BIDS format
• Review progress from development team
• Additional topics

Minutes

• Review of FHIR progress
  • Initial foray here ready for review
    • Make a comment (preferred) or email Nic
  • Brad to check with FHIR community
  • Bag of tags preferred; check out CodeableConcept
  • How should annotations be represented?
    • Eg like this?

[
  {original study = nifti}
    {annotation 1}
      {annotation 1 = nifti}
      {classification label}
    {annotation 2}
      {annotation 2 = nifti}
      {classification label}
    {annotation 3}
      {annotation 3 = nifti}
      {classification label}
 ]

• Flexible Configuration (e.g. check out NiftyNet issue )
  • Concept is zero-coding configuration to drive MONAI
    • right now, it takes minimum 5 lines of code
    • What workgroup should work on this?
      • Could be Data workgroup, could be Research workgroup
      • (Brad) take to steering committee
    • Data working group should endeavor for it to be zero-coding configuration, regardless
    • Support for online configuration management
      • Check out neptune.ai and mlflow
• Ivadomed
  • PyTorch library similar to MONAI also has a data loader; simple folder XML parsing mechanism
  • Stephen, Michael to reach out to community to explore synergy
• Fast track OpenData release
  • Working on AWS and open data registry, how to maintain
  • Data structure is highly relevant to Data working group
• Labels for Github issues
  • A nice recommendation is to have a tag for the working group
  • (Brad) to raise at steering committee