 |
 |
Hi Easiest is to have the white surface as seed and the pial surface as stop. Note: matrix1 will be quite large with freesurfer surfaces (~170K^2). Saad On 18 Apr 2012, at 14:28, Franz Liem wrote: > Dear Saad, > > I would like to use ptx2 to perform whole brain tractography with the freesurfer lh. & rh.white as seed and derive the connectivity matrix from each seedpoint to each seedpoint by using --omatrix1. > The problem with this is that tracing is continued after arriving at another point on the seed mask and seed and stop mask cannot be identical (also see https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1202&L=fsl&D=0&1=fsl&9=A&J=on&d=No+Match%3BMatch%3BMatches&z=4&P=344505). In this post you suggest perform N tractrographies with (N-1) targetmasks which forces one to define ROIs at the stage of ptx. I would rather stay a the "vertex space" in order to test different parcellation schemes after performing probtrackx. > > So my solution would be to perform N separate probtrackxs (N = total number of vertices) with modified white.asc files like this: > Lets just consider the lh for now. > N=1: seed surface file would be lh.white.asc with value at (1,4) = 1 > stop surface file would be lh.white.asc with value at (1,4)=0, (2:end,4)=1 (with end = end of vertex definition = no of Vertices; val(:,4) at the polygon definition = 0) > ... > Unfortunately, this approach yields only nonzero fdt path values (2x2x2 1mm voxels) around the seed point . (Without the stopmask, a clear tract is visible. Therefore, my surface definition seems to work.) > (I know that I will have to use omatrix2, but for now, I would just like to see whether the tractography might work.). > > Do you have an idea what the problem might be? > Alternatively, (and this might be asking for too much) is there an easy way to implement this (that is, allowing the seed and stop mask to be the same and in each run excluding the seedpoint from the stopmask) as a feature? I guess I would not be the only one appreciating it. Or, would you be so nice as to provide the probtrackx2 source code for me to have a try myself? I think this would also be much more advantages wrt. computation time, because the bedpostx samples only would be loaded once, not N times. > > Thank you so much for your help. > Franz > -- Saad Jbabdi University of Oxford, FMRIB Centre JR Hospital, Headington, OX3 9DU, UK (+44)1865-222466 (fax 717) www.fmrib.ox.ac.uk/~saad