CTSC:ARRA.110910

Home < CTSC:ARRA.110910

Agenda

Harvard Catalyst Medical Informatics group Meeting Minutes November 9, 2010

In attendance:

Testing of MGH laptop showed that the problems were due to wireless internet connection. After fixing that, Phase 3 was installed on the laptop by Bill with help from Darren and Ken Nesbitt. Testing showed that using the mi2b2 interface they could search for, find and download medical image files from the MGH PACS. Main problem was speed of download, going at 5 MB/minute. Bill Hanlon suggested checking network card connections on device, Darren will follow up on this. Next step is to do the Phase 3 install at BWH and CHB so that we can test whether the performance is also so slow before digging into other sources of the slowdown.

Another problem is that using the mi2b2 viewing software, we are not able to view the downloaded ultrasound images. Have some work arounds, but still looking into source for the problem and a more robust solution (e.g. swap ImageJ for another viewing tool).

Another more concerning issue is that for larger image data sets, downloads are incomplete- some images are missing. Not sure if this will be solved by improving the down load speed, but shows need to improve the 'safety net' to ensure complete data downloads.

Bill W and Ed Connely (sp?) will install Phase 3 at BWH this week and test with images identified under Randy's IRB.

Purchase of MGH server(s) has been on hold for final decision regarding issues raised in last week's meeting (e.g. do we need separate servers for mi2b2 and XNAT). Continued that discussion. Chris spoke with Tim Olsen (XNAT core developer) who says he does all his testing of XNAT on Windows so doesn't foresee a problem using a Windows OS base for the XNAT server. But see next note.

Group endorsed the provisional recommendation that for the next phase XNAT be on a separate server from mi2b2. Reasons for the recommendation are 1) that mi2b2 will use a Windows based OS and XNAT will use LINUX so that each is in the environment that allows the most robust support from the developers; 2) that mi2b2 test server is envisioned to live in the Radiology Departments for the near future, but the mi2b2 production server and the XNAT servers will live in the IT/IS Research Computing environments at most sites; 3) there is potential for port and other computer resource conflicts between mi2b2 and XNAT now or in the future that will be avoided by having them on separate machines; 4) more efficient use of resources if for instance Partners uses a single XNAT instance. Additional discussion to finalize this needed when we have more representation from BWH and CHB and when Jesse can let us know about BIDMC.

Once the recommendation is final, Randy will hold next meeting of Advisory Board to present this and get input.

Bill Hanlon suggests that duplicating the production instance of mi2b2 with a fail over server would be very important for supporting TIMC.

Discussed the need for keeping our cloud computing environments operational. Currently we have two instances up and running, one that Steve Pieper set up outside all firewalls and a second one Chris set up inside the MGH firewall. Steve suggested using the external one in the future for demonstrating the mi2b2 project.