Global tract-graphs

1

Dear experts,

I an trying run global tractography using the following commands;

5ttgen fsl T1.mif 5tt.mif

tckglobal dwi.mif wmr.txt -riso csfr.txt -riso gmr.txt -mask mask.mif -niter 1e9 -fod fod.mif -fiso fiso.mif tracks.tck

However I get the following error:

[ERROR] file “gm.txt” contains matrix, not vector

I would be grateful for your help.

Thanks

Peter

2

Hi Peter,

The file gmr.txt should be a vector of the same length as the number of b-values in the DWI image. It looks like your input file contains a matrix. Where did you get you response functions from?

Cheers,
Daan

3

Thanks Daan, I must have made an error before it seems to be working now. However the matrix I generate is very sparse, making me think there is still a problem. I have used the following commands;

dwi2response msmt_5tt DWI.mif 5tt.mif RF_WM.txt RF_GM.txt RF_CSF.txt -force

tckglobal DWI.mif RF_WM.txt -riso RF_CSF.txt -riso RF_GM.txt -mask nodif_brain_mask.nii -fod fod_glob.mif -fiso fiso.mif tracks_glob.tck -force

tcksift2 tracks_glob.tck fod_glob.mif sift_weights_glob.txt -force

tck2connectome tracks_glob.tck -tck_weights_in sift_weights_glob.txt diff_space_nodes_fixSGM.nii.gz connectome_glob.csv -zero -force

Thanks

Peter

4

Hi Peter,

At the very least, you’ll need to increase the number of iterations in tckglobal. For HCP data -niter 1e9 is usually sufficient.

If you need to further increase the density of the tractogram, you can decrease the particle weight by, say, a factor 5, while also multiplying the no. iterations by the same factor (i.e., -weight 0.02 -niter 5e9).

Cheers,
Daan

5

Thanks Daan, I have added these options, however while the matrix is slightly more populated it is still much more sparse than using tckgen based pipeline. Also it’s about an hr or longer for one subject.

Does this sound normal for tckglobal?

Thanks

Peter

6

Hi Peter,

Of course, the number of iterations will take a penalty on the run time. But it’s very important to give the algorithm time to converge. The default no. iterations in tckglobal is set to a small number suitable for testing in small test images or masks only, and should always be increased for full brain tractography.

Cheers,
Daan