Dear Gwenaelle

Thanks a lot, not it worked...

Coming back to the issue of multiple MRI scanners in VBM, do you have any good suggestions how to integrate this into the RANDOMISE analysis?

Thanks a lot

Sven

On 23 sept. 2011, at 17:02, Gwenaëlle DOUAUD wrote:

Hi Sven,

regarding multisite, I am not very familiar with these issues but I would actually believe that VBM is not that robust with different scanner/resolution. You could maybe get away with this if you have two populations in each site and are looking at the difference between these two: you will then need to pool the subjects by diagnosis and add a confounding regressor for the site...

You can re-run fslvbm2 without the first segmentation step by editing the fslvbm_2_template script:
put lines 36->45 in comments
and remove -j $fslvbm2a_id from line 52
this should hopefully work.

Hope this helps,
Gwenaelle
 

De : Sven Haller <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Jeudi 22 Septembre 2011 21h50
Objet : Re: [FSL] Re : [FSL] fslvbm_2_template does not create template_GM_4D

Dear Gwenaelle

Thanks a lot for your message

all fslvbm2?.e* are empty...

I checked again and I think that I might have found the problem. There were still filenames of deleted 3D Data in the template_list file (sorry, stupid error. I did modify this file, yet apparently it was not replaced, probably a permission problem as I am working on different computers....)

Also, how are you going to account for  the multi-site VBM? 
Good question. I do not know yet, I would like to see the data first. Generally speaking, you would hope that VBM is a generally applicable technique that provides robust results - yet the scanner problem should ideally be more an academic issue, correct??

Another question:
When I re-start fslvbm_2_template, the whole segmentation starts again from the beginning (yet it was already done...) Is there an easy way to avoid this re-calculation? I have 103 subjects, and each subject takes on my computer 10 min

Thanks
Sven



On 22.09.2011, at 20:14, Gwenaëlle DOUAUD wrote:

Hi Sven,

do you have any fslvbm2?.e* that is not empty in your struc directory?

Also, how are you going to account for  the multi-site VBM?

Cheers,
Gwenaelle

De : Sven Haller <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Jeudi 22 Septembre 2011 7h05
Objet : [FSL] fslvbm_2_template does not create template_GM_4D

Dear FSL experts

I try to do a VBM analysis in FSL

I have done several analyses in the past, without any problem

The only difference of this analysis is that this time the 3DT1 data are heterogeneous (different scanners and different resolutions)

The first step fslvbm_1_bet worked correctly, and I excluded all subjects with problematic BET

The second step fslvbm_2_template worked without error message, yet I did not get the  template_GM_4D (while I have a XXX_struc_GM.nii.gz for each subject).

I saw in the discussion list, that several people had problems before creating the template_GM_4D file, yet my situation seems a little different....

Any help is greatly appreciated !!

Sven

--------------------------------------------------------------------
Gwenaëlle Douaud, PhD
FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington OX3 9DU Oxford UK
Tel: +44 (0) 1865 222 523 Fax: +44 (0) 1865 222 717
--------------------------------------------------------------------