CTSC:ARRA.081710
From NAMIC Wiki
Revision as of 15:27, 17 August 2010 by RandyGollub (talk | contribs)
Home < CTSC:ARRA.081710
Back to CTSC:ARRA supplement
Agenda
- Phase 2 mi2b2 software update and discussion
- Presentation of GUI plans
Contents
Harvard Catalyst Medical Informatics group Meeting Minutes August 17, 2010
In attendance:
- Ron Kikinis
- Bill Wang
- Valerie Humblet
- Chris Herrick
- Charles McGow
- Shawn Murphy
- Steve Pieper
- Randy Gollub
- Bill Tellier
- Jon Bickel
- Kathy Andreiole (phone)
- Dan Marcus (phone)
- Alex Zaitsev
- Yong Gao
- Cynthia Lee
- Darren Sacks
- Jesse Wei
Meeting Minutes
mi2b2 software update and discussion
- Work this week focused on de-bugging display problems that appear to arise due to the use of ImageJ for visualization which doesn't deal well with compressed images. Solution ideas include adding decompression software on the mi2b2 server side. Per Kathy, every modality compresses the images 2-3:1 before it gets to the PACS. Steve suggests that if we build in a decompression function, we consider generating a Catalyst mi2b2 UID for every image retrieved- this is the convention for all DICOM image handling, when an image is changed. Dan says that in the XNAT de-identification tool box (DICOM editor) there is a function that decompresses the images and can optionally change their UID to an XNAT specific UID. Dan will share this software code and associated Toolboxes for us to use to incorporate into the mi2b2 code. This will cover the majority of the medical imaging data (~90-95% of cases). For the remainder, the specialty compression algorithms, we will flag the images as having "non-conforming compression". Non-conforming compression algorithms generally use the private fields (shadow fields) that are proprietary. Our mi2b2 DICOM conformance statement will include the information on our decompression methods, state we support this function for all shared compression schemes. Any images with proprietary or non-conforming compression will be provided "as is" and labeled as such. More fascinating discussion ensued regarding how to best implement the option of generating the unique mi2b2 UID.
Presentation of GUI plans
- Two options, "i2b2 aware" and "i2b2 unaware". In the "i2b2 unaware" situation, a user will begin with their own list of accession numbers or MRNs. Mi2b2 will search for available image data sets and generate a list of them. User can then select those that they desire, optional is use of filters to edit/select by display options (e.g. dates, gender, etc).