PS: oops! where I wrote z-stat I meant t-stat, the image that you are using with -mul in fslmaths. Regardless, you already got other insights from MJ!

Am 30/05/2014 23:58, schrieb Anderson M. Winkler:

Hi Xinfa,

The procedure to get the coordinates looks fine to me.

Regarding the difference, it might be that you are using the TFCE image in FSLview, whereas the peak of the clusters is being identified with the z-stat image; they aren't guaranteed to be the same, and will often differ: TFCE has the benefit if considering also the spatial extent, something the simple z-stat doesn't do.

All the best,

Anderson

On 29 May 2014 15:11, SUBSCRIBE FSL Xinfa_Shi <[log in to unmask]> wrote:

Hi ,

Now I have finished statistic analysis about FA data in TBSS by the command “randomise “, getting these files( tbss_tstat1.nii.gz , tbss_tstat2.nii.gz, tbss_tfce_corrp_tstat1.nii.gz , tbss_tfce_corrp_tstat2.nii.gz , tbss_tfce_p_tstat1.nii.gz , and tbss_tfce_p_tstat2.nii.gz) . Then I want to sort out the result following the steps below ：
1.      Getting Cluster and Peak Information from Randomise Output by following commands:
fslmaths tbss_tfce_corrp_tstat1 -thr 0.90 -bin -mul tbss_tstat1 tbss_thresh_tstat1

cluster --in=tbss_thresh_tstat1 --thresh=0.0001 --oindex=tbss_cluster_index --olmax=tbss_lmax.txt --osize=tbss_cluster_size

The result is :
Cluster Index   Voxels  Z-MAX Z-MAX X (mm)       Z-MAX Y (mm)       Z-MAX Z (mm)      Z-COG X (mm)       Z-COG Y (mm) Z-COG Z (mm)
6     2077    4.11     46  26  9     33.7     10.2     22.2
5     1190    4.68     -20 23  35  -22.3    29.2     18.4
4     850       3.55     19  3     41  15.8     17.6     32.8
3     104       3.99     21  46  3     18.9     41.3     -2.34
2     58  3.12     32  -34 13  30.9     -35.2    15.1
1     46  4.47     30  18  37  30.3     17.9     37.4

2.      View  tbss_tfce_corrp_tstat1 loading  the white matter atlas JHU-WhiteMatter-labels-1mm.nii.gz  in fslview by the following command:

fslview $FSLDIR/data/standard/MNI152_T1_1mm mean_FA_skeleton -l Green -b 0.2,0.7 tbss_tfce_corrp_tstat1 -l Red-Yellow -b 0.90,1
The result is seen in the attached picture .

Question:

1.      Is it right in the above process ?
2.      In FSLView , the coordinates (mm) presenting significant difference (Red-Yellow) are different from  the coordinates   (mm) deriving from cluster command .
That’s why ? What is the relation between them ?

Maybe my expression is not clear due to poor English ,and I’m very sorry for it .
In a word ,I want to know that how should I sort out the result presenting brain regions and clusters .

Best .

Xinfa Shi.