There are a couple of things that cause these asymmetries. 1) As Scott says,
one encounters a different pattern of local diffusion going in one direction
than one does going in the other.  2) This is more complex because newer
versions of FDT model 2 or more fibers.  Thus, tracts can become
non-symmetrical by choosing one fiber when tracking in one direction through
a voxel, but the other when tracking the other direction.  

Two mask symmetric actually does an A to B and adds it to a B to A, so one
can use that or do them separately.  I usually do them separately because
two mask symmetric didn't used to output a waytotal file (I don't know if it
still doesn't).  

Peace,

Matt.

-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Scott Kolbe
Sent: Tuesday, October 06, 2009 11:33 PM
To: [log in to unmask]
Subject: Re: [FSL] probtrackx conceptualization problem

Dear Dianne

I think about it in terms of /serial /sampling of the principal 
direction of diffusion (PDD). when you use probtrack with seed/waypoint 
your start sampling from all seed voxels and keep only the samples which 
pass through the waypoint. when you use two mask symmetric you seed from 
both regions and keep the samples which pass through the other (please 
check this as i can't seem to access the FSL website at the mom to check 
it myself). when you sample the PDD serially, errors will build up as a 
function of distance from the seed. so when you seed from both "ends" of 
a hypothesised tract you improve the SNR of your tract estimate.

In the case of cortico-cortico connections, I think it's best to track 
twice: first using a seed in region A and a waypoint in region B, then 
from B to A. The intersection of the two results is a good estimate of 
the tract location, and SNR should be reasonably good along the whole 
length of the pathway.

anyone please feel free to disagree.

cheers
Scott


Dianne Patterson wrote:
> Dear Group,
>
> I'm trying to wrap my head around a problem...and it just isn't 
> working. Perhaps someone can help reduce my confusion.
>
> If I am tracking between 2 masks, it seems to matter which is the seed 
> and which is the target.
> In fact, having a seed mask and a waypoint mask yields different 
> results than having 2 seed masks.
>
> There is one tract that I can get if I have the seed in the occipital 
> lobe and a waypoint in the anterior temporal lobe, 
> but if I use 2 seed rois, instead of a seed and waypoint, I don't 
> generate a tract.
>
> How can I make sense of the apparent assymetrical character of these 
> trackings?
> What does it mean for the underlying tracts?
>
> Thankyou for your time,
>
> Dianne
>
>
> -- 
> Dianne Patterson, Ph.D.
> [log in to unmask] <mailto:[log in to unmask]>
> University of Arizona
> SLHS 328
> 621-5105

-- 
========================
Scott Kolbe
Neuroimaging Group
Florey Neuroscience Institutes and
Centre for Neuroscience
University of Melbourne
VIC, Australia, 3010.

ph:       +61 3 8344 1929
email:    [log in to unmask]
website:  www.neuroimaging.org.au/index.php?id=383