Hi Chiara,

Hmm, that really should not happen. You see, the fslvbm2 script does segment with FAST *all* of the images, regardless of what is in the template_list file, so you should have at the very minimum a *_struc_GM.nii.gz for all of your 48 subjects. Could you please type this in your struc directory and tell me what you get?

ls *_struc_GM.nii.gz | wc -w

Cheers,Gwenaëlle
 
------------------------------------------------------------------------------------------------------------------------
Gwenaëlle Douaud, PhD
Associate Professor & MRC Career Development Fellow

FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington
OX3 9DU Oxford UK
Switchboard: +44 (0) 1865 222 493
Fax: +44 (0) 1865 222 717

www.fmrib.ox.ac.uk/team/principal-investigators/gwenaelle-douaud
www.fmrib.ox.ac.uk/research/translational-image-analysis-group
-----------------------------------------------------------------------------------------------------------------------


De : Chiara Moda <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Lundi 5 décembre 2016 17h00
Objet : Re: [FSL] Add subjects to VBM analysis

Hi Gwenaelle, 

Thanks for your reply. 

I was initially doing as you suggest: keeping all scans in one directory and creating the template_list with the group sub-sets. (as opposed to adding them later..I thought this may help) but I still have the same problem.

When I run fslvbm_3_proc it only ever takes the scans that have been named in the template_list to create the modulated image. So, if I load GM_mod_merg_s3.nii.gz it will only ever have the 30 volumes from the template and not the 48 subjects in total despite them being in the directory. Of course, I always get an error that the design matrix does not have the correct number of data points. 

Going through my 'struc' folder just now I have noticed that some of the scans not included in the template only have the following files created:

*_struc_brain.nii.gz
*_struc_GM_to_template_GM_warp.msf
*_struc.nii.gz

I am not sure if any of this points to an obvious problem, but I have had this issue for several months!

Many thanks again,

BW

Chiara


On Mon, Dec 5, 2016 at 4:36 PM, Gwenaëlle DOUAUD <[log in to unmask]> wrote:
Hi Chiara,

You should not have to do all this. Include *all* of the scans for your VBM processing steps. The only step to match the size of the groups is for the creation of the template, and this can simply be done via the selection of the right number of scans (for the two bigger groups, you will have to randomly draw the same number as in your smallest group) in the template_list file.

Cheers,Gwenaëlle
 
------------------------------ ------------------------------ ------------------------------ ------------------------------
Gwenaëlle Douaud, PhD
Associate Professor & MRC Career Development Fellow

FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington
OX3 9DU Oxford UK
Switchboard: +44 (0) 1865 222 493
Fax: +44 (0) 1865 222 717

www.fmrib.ox.ac.uk/team/ principal-investigators/ gwenaelle-douaud
www.fmrib.ox.ac.uk/research/ translational-image-analysis- group
------------------------------ ------------------------------ ------------------------------ -----------------------------


De : Chiara Moda <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Lundi 5 décembre 2016 12h35
Objet : [FSL] Add subjects to VBM analysis

Hello, 

I am running VBM to compare three groups (two patient, one control). 

My groups are unequal sizes so I am using a subset of two groups to match the smallest group. I want to add the additional subjects back in for the fslvbm_3_proc command. 

So, I have added the additional scans to the directory and ran flvbm_1_bet on them. I have created the design matrix and contrast files with all subjects included. 

When I run fslvbm_3_proc, a list of all subjects appears in the terminal but, only the subjects included in the template are actually included the GM_mod_merg* files. 

Have I missed a step? Could someone help me understand why my steps are not letting the command know that I have added subjects so I can run randomise?

Thank you very much in advance!

Chiara