When I try to load trk files with b0 images, the two do not overlay perfectly until I manually set the nifti orientation to the nearest diagonal matrix. For example, the following nifti header was dumped from the output b0 image after recreating tracts from dicom:

http://paste.pocoo.org/show/443305/

When I actually load the image with the trk points, they don't overlap until I set the orientation to:

http://paste.pocoo.org/show/443312/

What coordinate system are the points in the track file supposed to be aligned to? Thanks for the help. I don't know if this is a bug in DTK or not, so sorry if this is a miss-post.

Post edited 11:34 pm – July 25, 2011 by yiuin98

Pixel space with the spacing information intact? 

In my opinion ignoring oblique information is a bug. What if you load a T1 anatomical image that has different oblique information? The two images won't show up correctly. Considering that DTI images take a long time to gather, gathering in a different orientation from the T1 anatomical is very likely to minimize scanner time. 

Yes, but it is a larger deformation registration than would be necessary if the two images had nearby orientation (assuming you use a registration algorithm that uses oblique information). Besides registration doesn't work for flips. Am I the only person who doesn't think we should throw out information?