Hey Yaeji,

this is a simple script that extracts the mean FA value from a standard FSL white matter atlas and writes it into a table per subject for further statistical analysis. You have to change the numbers according to the name of the subjects folder in your tbss stats folder and the file path of you FSL atlases.

#! /bin/bash

# preproc for roinum in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 ; do

fslmaths /usr/local/fsl/data/atlases/JHU/JHU-ICBM-labels-1mm.nii.gz -thr $roinum -uthr $roinum -bin roimask

fslmaths roimask -mas mean_FA_skeleton_mask.nii.gz -bin

roimask

padroi=`$FSLDIR/bin/zeropad $roinum 3`

fslmeants -i all_FA_skeletonised.nii.gz -m roimask -o meants_roi${padroi}.txt done

for extracting mean diffusity check this: http://wikis.la.utexas.edu/imagelab/book/statistical-analysis-mean-diffusivity-md.html

Good luck,

Hans

Have you carried out any preprocessing other than eddy? You should first denoise, remove Gibbs artefacts, and correct for bias, and then run eddy. If you don't run any of these corrections before eddy, then you are using data that may have noise which could potentially obscure any significant results (if there are any). What type of data do you have e.g. 3T, spatial resolution, slice thickness, b-value? These all impact upon SNR.

Which version of eddy are you running? e.g. eddy_openmp, eddy_cuda ? eddy_cuda has an option of slice-to-volume correction instead of just volume-to-volume correction.

Do you have blipped data as well, or only 1 phase-encoded direction of data?

TBSS only looks at the WM skeleton, so you may want to consider dMRI (DTI) VBM, or tractography of specific tracks and compare the FA, MD etc of each of those tracks between your 3 groups.

Also, keep in mind that FA is a conglomerate of L1, L2, and L3, so you are more likely to find differences between groups in L1, L2, and/or L3 rather than FA. That is, make sure to compare RD and AD between groups, and not just FA and MD.

Jerome

I think you can try 3 d slicer software from slicer.org

Hans-Christoph Aster Thank you! I will try those script

Jerome Maller Thank you for the kind response. I simply run the eddy_correct , not eddy_cuba or eddy_oepnmp. Should I do those options, other than eddy_correct?

To denoise and to remove Gibbs artifacts, what tool or script should I use?

Also I am considering to use DSI studio for structural connectivity, but do you recommend any easy tool? Because I am very new to NeuroImaging world.

Thank you so much!

eddy_openmp will utilise all of your CPU cores. So, if you have a CPU with 8 cores, eddy_openmp will use them all instead of just 1 (which is eddy_correct).

If you have CUDA installed, then eddy_cuda will use the graphics card and will be much quicker than even eddy_openmp.

Use MRtrix3 to denoise:

dwidenoise dti.nii denoised.mif -noise noise.mif

The input is dti.nii, the output is denoised.mif, and the noise map (if you are interested in looking at it) is noise.mif

Use MRtrix3 to remove Gibbs artifacts:

mrdegibbs denoised.mif denoised_gibbs.mif

The input is denoised.mif, the output is denoised_gibbs.mif

-Jerome

Hans-Christoph Aster can the same be done but instead extracting the FA from a standard atlas using specific white matter pathways tracked by probabilistic tractography?

Dragana Galevska Sure, you can use the script to use any standard Atlas with labels as a mask for your FA data. In this case the labels should be numbers (1,2,3..) in my atlas this goes up to 48 labels.

Hi everyone,

I have one question if I may :

- Is it correct to extract mean FA values from a tbss skeleton using a tractography based atlas like this one ==> (http://www.bcblab.com/BCB/Atlas_of_Human_Brain_Connections.html) ?

Thank you !