Puget Sound Collaboration


Jump to: navigation, search


[edit] March 16 2010 Meeting

[edit] Proposed agenda

In anticipation of our meeting tomorrow, I put together some brief notes. No need to respond to this before the meeting - but hopefully you'll have a chance to skim this e-mail briefly:

Ideas for the proposal:

I've got a long list of ideas from Daniel Glen at NIH, which I will forward on to the group separately.

A few top items on my own personal list include:

1) Completing the integration of DTI-Query and CINCH. Tony et al. have come a long way on this, but it needs to be finished and released to the community.

2) Better support/advice for users who need help transforming their raw DWI data to be used with DTI-Query. This is by far the most commonly-asked question that I get. Besides multi-platform support, this may be the biggest impediment to adoption. We could just tell them to use DTI-Query together with the VISTA suite, but many groups already have their own software setup, and/or don't have Matlab.

3) Better multi-platform support. As far as I know, DTI-Query is the only software of ours that works on the Mac platform. Doing this port immediately had a drastic effect on the number of downloads. It may not be exciting from a computer science standpoint, but it makes a big difference to adoption.

4) A new website to support the software... (We've already talked about the need for this.)

What I'm looking for:

My initial thoughts are to ask for salary for me and 1-2 students, for one or two summers. (At my institution, teaching responsibilities make it difficult to find the time for research at other times during the year.) This would likely amount to about $25,000-$30,000 per summer in direct costs. I'm not sure what the indirect costs are at UPS, but I expect somewhere around 40%. I would also plan to ask for travel costs between here and Stanford, so that we could meet. Equipment costs should be negligible - maybe a few thousand dollars?

Which grants to target:

Daniel mentioned that we would be MUCH more likely to get most grants if we can show a collaboration between our institutions. There are literally dozens of options, but the one that seems the closest to our interests is this one:


It's about neuroscience software modification and enhancement that are included or considered for inclusion into the NITRC website. However, unfortunately the latest date for submission listed is May 15 (letter of intent due April 14!). I suspect that we might not have time to put together a proposal by then without a big push.

Here are some other miscellaneous options, sent to me by Daniel at NIH. Some of them are pretty vague, but we can talk over which ones might be best (and I will rely on your experience here, since I have never submitted a grant to NIH):




(Brian's preference) http://grants.nih.gov/grants/guide/pa-files/PAR-09-218.html
















[edit] Meeting notes

Common interests: getting Quench out there and maintaining it

Database integration? Plug-in architecture so that we can support multiple databases?

For Dave, the RO3 might be a better mechanism than the RO1. Dave will look into the RO3 application effort. It is due by May 14.

[edit] Ideas from Dan Glen

A message from Daniel Glen at NIH. One thing I left out in the previous e-mail (but is obviously of interest to everyone) is to integrate better tractography solutions with the visualization tools that we ship. Obviously, the three of you have a lot to contribute there.

Hi Dave,

I haven't tried CINCH, but the biggest piece missing from DTI-Query is the easy input of irregular VOIs (not just cubes). Mike Beauchamp and Tim Ellmore have gotten around this partially by creating a lot of small cubical VOIs (which is basically what it would be in the end anyway), but that makes the query logic difficult. You (and I helped too) changed the input mask VOI be a seed mask and not a limit mask, but both are needed. I think this request was one of the first I made, and it has been repeated by several others. Daniel Handwerker here says he has had problems when he has used high-resolution volumes 1mm3, so that's an issue. I would also make the request to do more via command-line interface - creating the fiber volumes (which is a really important option for effectively getting to a group analysis).

I think none of the other tractography packages really approach the interactivity yet of DTI-Query, but software can change quickly. You should probably be more pro-active in making users aware they can drag the VOI around because that is really the best feature. Mike Beauchamp tells me it fits the general philosophy of ease of use in data exploration already followed in AFNI. AFNI's graphing, multiple viewers, interactive clustering and thresholding, instant temporal correlation and InstaCalc (instant image calculation) are some of these features that have that same kind of instant feel that DTI-Query has. Larry Frank's group from UCSD has been working on a Diffusion plug-in for AFNI which is more integrated into AFNI, but it suffers from a fair amount of bugginess and the tractography algorithms are not working too well so far. One of their tractography algorithms is based on a seed-based water diffusion model and can handle fiber crossing well (at least theoretically). Right now TrackVis is the most directly comparable package, and it works reasonably well. Though it has a slightly quirky interface, it is flexible regarding ROI/VOI input. DTI-Studio provides a full pipeline (well, not group analysis) and is probably the most widely used DTI package. Ideas about group analysis including alignment to standard templates based on major tracts could be useful and proposal of standard methods for the statistical analysis could be useful. Other ideas, though not as important, that I can think of:

1. Multiple views for tensor, eigenvector, FA, FA-encoded primary eigenvector 2. Opening linked multiple viewers 3. FA-encoded fiber view

I am sure if I think about it, I can come up with more.

I think it's going to be difficult to start off on your own without collaboration from either Stanford or some other group in the region (University of Washington or VA Puget Sound) to get funding, but perhaps one of the young investigator type awards would work.

Regards, Daniel

Personal tools