Hi Saad and Longchuan,
I've observed same strange output coordinates in my side, using matrix1
mode and combined ascii surfaces and volumes seeds.
And Longchuan,
You're right about surfaces indices in fdt_matrix3.dot, but I think they
start at one and not zero. You can also use spconvert (matlab) to easily
convert into connectivity matrix.
Arno,
Hi, Saad
Thank you for checking. The exact command I used for loading into matlab was :
M = read_avw('coords_for_fdt_matrix1.nii.gz').
I ran the probtrackx2 (build 500) twice and the results are the same. It is quite strange.
I also noticed that the fdt_paths.nii.gz results between the two versions are identical (build 500 vs. build 4110), at least based on the one-sample-per-vertex test I did. So if the cause of the strange coordinates cannot be diagnosed shortly, do you recommend running the probtrackx2 with the build 4110? I just want to make sure there are not other major modifications between the two versions that may give me less accurate results.
Another related question is fdt_omatrix1.dot. Am I correct that the first and second columns are the indices of the two vertices that are connected, and these indices start from zero? if one vertex in the left hemisphere (the first seed surface mask in the text list) is connected to a vertex in the right hemisphere (the second seed surface in the text list), should the first, second and third column of fdt_omatrix1.dot for that connection be:
first vertex index second vertex index + total vertex number in the first seed surface number of connections
Sorry for so many questions and I really appreciate your help!
Longchuan
From: Saad Jbabdi <[log in to unmask]>
To: [log in to unmask]
Sent: Monday, October 15, 2012 4:54 PM
Subject: Re: [FSL] a question on probtrackx2 (build 500) with surfaces
Hi - Sorry I can't reproduce your result.
Can you send the exact syntax you are using in matlab for loading and looking through the coords_for_fdt_matrix1.nii.gz file?
You are right in your interpretation of columns 4 and 5 :)
Cheers
Saad
On 15 Oct 2012, at 20:01, Longchuan Li wrote:
> Hi, Saad and other FSL experts
>
> I have a question regarding the "coords_for_fdt_matrix1.nii.gz" file generated using probtrackx2 (build 500) and would appreciate your help on it.
>
> I find that the file generated by probtrackx2 (build 500) looks strange. It is N*5 matrix with all five columns like some coordinate values. I opened it using read_avw and extracted the first three rows for example.
>
> -6.53106021881104 -25.1907997131348 -12.3368997573853 -42.0928993225098 -48.9939002990723
> -6.79067993164063 -26.1084003448486 -15.0829000473022 -44.4836006164551 -49.4682006835938
> -7.44923019409180 -28.8355998992920 -15.3670997619629 -45.2563018798828 -50.6940002441406
>
>
> As I downloaded the patch before the formal release, I also have the probtrackx2 (build 4110) and used the identical command line. I got a file named "coords_for_fdt_matrix1" with the contents as follows:
>
> -6.531060219 -89.15630341 -11.68620014 0 0
> -6.790679932 -89.26719666 -11.9052 0 1
> -7.449230194 -89.165802 -11.98849964 0 2
>
>
> These results make much more sense to me. I assume that the first three columns are the coordinates of the vertices and the *fourth* column is the index of the seed surfaces ( I have two seed surfaces i.e., left and right white matter surfaces with medial wall excluded).
>
> Do you know why the newer version has that strange format?
>
> Below is the command line I used for tractography:
>
> probtrackx2 --seed=${args[1]}/${args[2]}/label/SeedList.txt --omatrix1 -l -S 2000 --xfm=data/xfms/fs2str2diff.mat --steplength=${steplength} -P ${samp_vox} ${distthresh_mode} --stop=label/StopList.txt --forcedir --opd -s data.bedpostX/merged -m data/nodif_brain_mask --dir=probtrackx/output_omatrix${args[5]} --meshspace=freesurfer --seedref=mri/orig.nii.gz
>
> Many thanks in advance!
>
> Longchuan
>
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
|