CTSC:ARRA.032911

Back to CTSC:ARRA supplement

Harvard Catalyst Medical Informatics group Meeting Minutes March 29, 2011

In attendance:

• Bill Wang
• Valerie Humblet
• Yong Gao
• Chris Herrick
• Shawn Murphy
• Steve Pieper
• Bill Tellier
• Darren Sack
• Bill Hanlon
• Alex Zeitsev
• Ron Kikinis

mi2b2 software update

• Bill recapped what was said last week about MGH: As the team worked on the multi-node, multi-thread workflow programming for the MGH deployment they ran into an issue where the same studies were found on more than one node. It turns out that there are multiple gateway nodes into which the various imaging data sets enter the disk storage systems. But once the data has been entered into the permanent archival system, queries to a single node should reveal all studies.

• At MGH, C-find still not able to move more than 2000 images or series. It is a problem with the configuration of the PACS.

• BWH: Kathy prefers to put the machine at the Crosstown facility because it is where they have all the research servers. Right now they are going through some big updates so mi2b2 is not a priority for them right now. Shawn prefers to have it at Needham because it will be more convenient for maintenance in the long run.

• Study reports embedded are not being download. The DICOM structured documents are treated as a separate series.

• Configuration of infrastructure for multi-node, multi-institution in progress.

• Last week the concern that was raised was how to control what machine is being used to run the software and is thus downloading the data to that machine. Partners does not require such a thing once people have their IRB approved so we should not spend a lot of time trying to build something not needed. Just allowing people to specify location where they will move the data is fine (example: encrypted laptop, computer behind the firewall). We can allow people to have limited data sets on their encrypted laptop for display. If performance is a concern as alternative, we first put the images on the server and then we send them to the client and they don't need to cache the images. Maybe we can give clients permission to use the mi2b2 cache. Part of the issue is that for now we zip all the files, think about another process.