NIH | National Cancer Institute | NCI Wiki  

Agenda/Notes:

  • Qinyan demo'd tool for converting LIDC to AIM
    • Notes on LIDC conversion process:
      • Run LIDC --> AIM conversion against the NCIA downloaded XML
      • Run MAX against the NCIA downloaded XML
      • Run LIDC --> AIM conversation against the MAX'd XML to create annotation of annotation
  • Discussed significance of "DateRequest" and "TimeRequest" fields in original XML
    • convert this to something in AIM?  or just scrap it?
    • Sam: there was also a "DateService" field but this is not in every XML
    • the most important of any of this is the DateRequest - this is the time that the XML was completed by the LIDC group
    • This information will be stored in the datetime element of the ? class (Qinyan will find out).
  • AIM requires a study date field
    • this information is not in the original XMLs
    • since we don't have this information our options are to use an obviously false date or pull it from the dicom
    • consensus on the call was to use a reference date
    • if possible can we use a format like "99-99-9999" to satisfy the format but have it clearly show it's not a real date?
    • Is the Study Date in the AIM annotation required to accurately represent the image with its annotation in an AIM-enabled browser like IPAD/Osirix or AIM-ClearCanvas?
  • Prioritization of remaining tasks
    • NCIA team has put together a tentative project plan
    • John/Justin will review with Qinyan to help prioritize tasks and ensure plan makes sense
  • Validating the conversion upon completion
    • we could use some of these AIM compliant viewers to visualize the AIM and see if the markup is redrawn properly
    • John mentioned the possibility of using Northwestern's ClearCanvas plugin once it's ready
    • Mike mentioned we might even be able to do this with AVT - should coordinate with Bob Schwanke
    • for validating the other information (not drawn marks) Sam has a lot of additional information about the cases in batch 1 that might help us confirm accuracy
    • Mike said that visualizing side-by-side the original in QIWS and the AIM versions in an AIM-enabled browser would be necessary
  • No labels