Hi Omar,

I'm glad you found the blog post helpful. Before I get to your question, I just wanted to point out two things. First is that in your 3dcalc -expr, you're essentially setting a threshold (here 0.25) which isn't mandatory, but when I wrote the blog post I was referring to one of Daniel Glen's posts. Second, I have found that using TORTOISE for preprocessing may yield slightly better results than the instructions on the blog using 3dAllineate. I have written instructions for using TORTOISE as well, based on the instructions, some correspondences, and my own fiddling:

Part 1 - [blog.cogneurostats.com]
Part 2 - [blog.cogneurostats.com]
Part 3 - [blog.cogneurostats.com]

To your questions (and Paul Taylor may have more to add), also I'm typing this as I hop from meeting to meeting today, so I reserve the right to go back and edit later on:

1. I would like to perform probabilistic tractography, but can it only be done between certain points or for the whole brain?

Usually tractography is done with seed regions. In the case of 3dTrackID, you get a metric of how many tracks run through a particular ROI or connect two ROIs. If you just want visualizations you can use the -logic OR and place seeds in the CC.

2. Also, after running the scripts above, is it convenient to warp the images to MNI if I want to select some ROIs to run tractography for group comparisons, just like fMRI? and how would I warp these images?

My current approach is to use group-level ROIs (anatomical or functional), warp them back to subject space using 3dfractionize and then fit the tensors via 3dTrackID. I then use the metrics for number of tracts within an ROI or between ROIs.

Approaches I haven't tried extensively: 1) You could warp tracks fit at the subject level to MNI (or other space) using map_TrackID. 2) If you use TORTOISE (DIFF_PREP), you can specify a template image (like Colin27 or MNI152) and it will affine warp the DWI images to the template and then using DIFF_CALC you can fit the tensors in that space. Your mileage may vary...

3. Also, is it convenient to run 3dDWUncert first and then 3dTrackID -mode PROB?

Yes. Checkout the tutorials that Paul Taylor put together by running the script @Install_FATCAT_DEMO on your computer. It will download some data and scripts that can be modified for your data.



Edited 1 time(s). Last edit at 10/16/2014 05:03PM by Peter Molfese.