Thank you so much for your response Saad.
Just to make sure I understand..If I'm not using distance correction, how many streamlines are typically sent from a seed voxel to the target? If I set my # of streamlines to be 5000, does each of the target get 5000 streamlines from the seed (which makes the total no.of streamlines 5000*no.of targets) , or is it just 5000 streamlines per seed voxel (and does not depend on the # of targets used) ?
Thanks again,
Priti
Hi
> First:
> I used the default # of samples (5000). But when I check my seeds_to_${target} file, majority of voxels have a huge number (in the order of 100,000 - 300,000). Hence I concluded that the total no. of samples sent out from any voxel in the seed mask = No.of samples * No.of targets. Please let me know if this assumption is right.
This is because you have the "distance correction" turned on, so it is multiplying the number of counts by the average distance. Turn it off to get number of samples.
>
> Second:
> I have 'n' targets that I've used for this analysis. If I have a streamline/sample that reaches target 1 and then goes on to target 2, will that streamline be counted in both places, will it be counted in the target2 or will it be counted in target1? The reason for asking this question is that I found a lot of streamlines connected to regions that are very close to my seed but that is not the case with regions farther from the seed, in-spite of doing path distance correction. Is there a way to get just the # of streamlines that terminate in region without counting the no.of streamlines that pass through that region and terminate elsewhere?
You can set all your targets as stopping masks (add them using fslmaths and set the result as stopping mask), so it only counts samples that go directly to each target without first crossing another one.
Cheers
Saad
>
> Thank you so much for all your help and time,
>
> -Priti
>
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
|