I am not so concerned with improving results, but avoiding the upsampling of the data from 1mm to 2mm.  As well I don't understand why the script runs without errors but fails to output necessary files, basically why the fnirt warping and apply warp steps in fsl_reg (2nd pass) don't seem to run and generate no errors.  I will keep trying solutions. 

Thank you for all your help I will read the posts you have suggested.

Much appreciated.

Joseph Ackerman Jr
Cell: 314-623-2605
From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Gwenaëlle DOUAUD [[log in to unmask]]
Sent: Friday, May 10, 2013 12:11 PM
To: [log in to unmask]
Subject: Re: [FSL] fslvbm_2_template -n with custom template

Hi Joseph,

as I said in my previous email, I'm afraid that getting a higher resolution for the template (1mm) would not change/improve the results. This has to do with the fnirt warp resolution -- see previous posts by Jesper on this by entering his name and 1mm as keyword:

https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1111&L=FSL&P=R50226&I=-3&d=No+Match%3BMatch%3BMatches
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1203&L=FSL&P=R6144&I=-3&d=No+Match%3BMatch%3BMatches

Cheers,
Gwenaëlle
 
------------------------------------------------------------------------------------
Gwenaëlle Douaud, PhD
MRC Career Development Fellow

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

www.ndcn.ox.ac.uk/team/researchers/gwenaelle-douaud
-----------------------------------------------------------------------------------

De : "Ackerman, JosephJr." <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Jeudi 9 mai 2013 22h40
Objet : Re: [FSL] fslvbm_2_template -n with custom template

I have attempted your suggestions and everything seem to work well.  One issue is that the template_GM (cohort specific) created by fslvbm_2 is 2x2x2mm and our t2w data is 1x1x1.  Thus when registering the individual brains to the template they will be upscaled to 2x2x2.  I would prefer to avoid this loss of resolution.  The two solutions I can think of are: (1) use pur cohort specific grey matter template so the input and reference space are the same resolution and our processing stays in 1x1x1mm.  (2) a technique I used before when correcting diffusion data (1.2 isotrop dti corrected with 1mm isotrop T2W) with flirt, fnirt and applywarp which was to use the input image of applywarp as its own reference and use the inverted (converted with convert_xfm) affine matrix from flirt as the –postmat after applying the warp from fnirt registration to put the image back into its own space (resolution) but with the correction from the fnirt warp.
 
I tried changing the Template in fslvbm_2 and it causes the 2nd pass fsl_reg to fail (without any errors but files not produced).  I also tried to edit the fsl_reg script (a copy I made) to change the fnirt option and the following applywarp step to mimic my 2nd idea above.  This again caused no errors but the files generated are different from what I get using the un-touched fslvbm_2.  In both the template_GM_init and Template_GM, which you identified as the two to check against each other (thank you for that) are identical.  It is as if the second pass (fsl_reg) is not being done.  
 
The easiest fix I can think of is to change the template called for by fslvbm_2 but that has yet to work for me.  Is there some special form / type of file that the template must be?  I see in the /standard/tissuepriors directory that it is .hdr and .img format (Analyze?).  I have tried my template as both Analyze. .nii and .nii.gz formats with identical failures. 
 
I am still searching the message boards for potential solutions but have not noticed anyone posting a resolution to the type of problem.  Other issues near this appear to output error messages of which I receive none. 
 
Thank you for any and all help.
 
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Gwenaëlle DOUAUD
Sent: Thursday, May 02, 2013 12:38 PM
To: [log in to unmask]
Subject: Re: [FSL] fslvbm_2_template -n with custom template
 
Hi Joseph,
 
I see... I haven't got much experience with such young subjects (before the intensity flip). If you are happy with the grey matter segmentation on your T2-w, and with your FSL-VBM study-specific template (check the final, 2nd pass one: template_GM versus the affine, 1st pass one: template_GM_init) then I'd say, use the default FSL-VBM settings... Getting a higher resolution for the template (1mm) would not change the results (see previous posts by Jesper on this).
 
Hope this helps,
Gwenaëlle
 
--------------------------------------------------------------------
Gwenaëlle Douaud, PhD
MRC Career Development Fellow

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

www.ndcn.ox.ac.uk/team/researchers/gwenaelle-douaud
--------------------------------------------------------------------
De : "Ackerman, JosephJr." <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Jeudi 2 mai 2013 18h18
Objet : Re: [FSL] fslvbm_2_template -n with custom template
 
My data is babies and T2W images since the ages are prior to the grey-white intensity flip.  My group felt using a grey matter template generated from similar aged brains (which we have) would give better results than starting with a registration to the adult grey segmentation.  As well our data is 1x1x1 as is our average grey segmentation while the adult template is 2mm.  As you are a more experienced user than I, if you feel that using the standard adult t1w template won’t negatively influence the template creation then that is great news. 
 
Since it is used for only an affine “rough” registration estimation I can see how it may not influence final out comes.
 
Thanks for your help in this matter.
 
Joseph
 
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Gwenaëlle DOUAUD
Sent: Thursday, May 02, 2013 12:05 PM
To: [log in to unmask]
Subject: Re: [FSL] fslvbm_2_template -n with custom template
 
Hi Joseph,
 
just one question before delving more specifically (if needed) in your issues: why do you want to use this "custom grey matter template", when FSL-VBM precisely create such study-specific template (in a two step fashion: first by registering all your GM images to the MNI152 GM, and flipping/averaging them, second by non-linearly re-registering/flipping/averaging all your GM images to this affine study-specific first-pass template)?
 
Cheers,
Gwenaëlle
--------------------------------------------------------------------
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
www.fmrib.ox.ac.uk/~douaud
--------------------------------------------------------------------
De : Joseph Ackerman <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Mercredi 1 mai 2013 20h54
Objet : [FSL] fslvbm_2_template -n with custom template

Hello.  I am attempting to run fslvbm_2_template using a custom grey matter template built for my cohort of subjects.  When I run the fslvbm steps without any alterations everything works fine, ie outputs are correctly generated.  My population is younger and thus I am trying to use a custom general template (prior) and atlas.  When I change the T= field from fsl/standard to the location of my template (which is .nii format) fslvbm_2 runs without any errors generated but there are files missing in the final /struc directory.  I tried fslchfiletype to switch my template to analyze as the provided template was in .hdr .img format with no change in results. The missing files are:
   
  *_GM_to_template_GM_init.log
  *_GM_to_T_init_warp.nii.gz
  *_GM_to_T_init.nii.gz (this file began showing up once I altered the $REG variable in the echo fsl_reg (nonlinear) to -a)

all the template_* files are created (probably from the first fsl_reg step in vbm_2 and I checked their alignment.  The alignments are good and  the _flipped is indeed flipped on the x axis but I don't think the program is running the fnirt process under fsl_reg.

Originally I was using a cohort of 6 controls and 6 others, but I have reduced that to 1 and 1 in order to decrease run times for testing.  The  segmentations from fast in vbm_2_template look good too. 
Any ideas why the -n (nonlinear) step of fsl_reg would be failing in fslvbm_2_template and not produce any error message outputs?  I read many similar questions on the message board but could not find any with a resolution that reflected the problems I am having (mainly because I don't get any error messages).

Thanks for any and all help.

Joseph
 

The material in this message is private and may contain Protected Healthcare Information (PHI). If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.