-
Monai Label Workshop this week
-
https://catalog.us-east-1.prod.workshops.aws/workshops/ff6964ec-b880-45d4-bc1e-468b0c7fa854/en-US/
-
https://github.com/NA-MIC/ProjectWeek/blob/master/PW37_2022_Virtual/MONAILabel_Workshop.md
-
DICOM Max Limits (real world)
- Largest ImageFrame (width/height/bpp)
- DICOM Standard max is unsighed short width/height (65535x65535) - never seen an image this big in the real world
- Digital Pathology images can be as large as 300,000x300,000 but these are tiled in DICOM to fit the 65535x65535 size
- https://dicom.nema.org/dicom/dicomwsi/
- Tend to be small 256x256 (JPEG baseline encoded)
- https://gist.github.com/PantelisGeorgiadis/fb9e305610c92633fd93fcc3e973d8a2
- Width of 16,000
- Largest Study (MB)
- Digital Pathology WSI 40GB
- Rumors of sample pathology studies in terabyte range
- Multiple wave lengths
- Multiple focal depths
- Million pixels on one side
- Largest Number of Instances / Study
- 65,000 fMRI
- Largest SOP Instance (MB)
- 40GB - Video (long running surgical video)
- Largest number of frames/sop instance
- 340,000 for Digital Pathology / WSI
- DCM4CHE/Orthanc take 30-40 seconds to retrieve last few frames
- 340,000 for Digital Pathology / WSI
- Largest clinical archive size
- 23 PB
- Largest ImageFrame (width/height/bpp)
-
HTJ2K Cornerstone Integration Update
- Project week project
- dicomParser needs a fix