Hi Irina

I believe Matthew recently provided some solution to this exact problem for someone else on the list. You can try that solution for now.
But if your FA folder contains 20000+ images that ends with _FA.nii.gz, not sure if that solution would work. Still worth a try.

I'm puzzled that your folder contains that many images that ends with FA ending - it should have appended _warp postfix to majority of the files.

I'm away from my computer right now but when I get a chance I can suggest a better solution for you.

Best

Danny Kim
________________________________________
From: FSL - FMRIB's Software Library [[log in to unmask]] On Behalf Of Irina Anurova [[log in to unmask]]
Sent: Wednesday, December 16, 2015 7:03 PM
To: [log in to unmask]
Subject: Re: [FSL] non-FA TBSS error

Hi Danny,
Thank you very much for helping me. You are right, there are 164x164=26 896 *._FA_FA.nii.gz files inside the FA folder, because the -n option coregisteres each subject's image with all other images from the group (164 subjects totally) in order to identify the "ideal" representative subject. What do you think, would it help if I remove all the images from the FA folder except for those which a coregistered with the "ideal" one?

On Thu, Dec 17, 2015 at 3:14 AM, Daniel Kim <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Hi Irina,

You should be able to use tbss_2_reg -n or -T results for non-FA TBSS.

Reason why I asked you to look in the FA folder is because of the error message you posted earlier: "imglob: Argument list too long"

In the tbss_non_FA script, one of the first thing that it does is get all the names of the subject images using imglob.

Under the FA folder, imglob *_FA.* should only list 164 images that you have, but I have a feeling that for whatever reason imglob is finding way too many images.

Are there more images inside FA folder that ends with _FA.nii.gz?

Danny Kim

On 2015-12-16, at 4:05 PM, Irina Anurova wrote:

Thank you very much, Danny (sorry, I've mixed up your name in the previous e-mail). I'll try tomorrow.
I found several posts almost exactly like mine on this forum, and sometimes people said that they have used the -n option in the FA TBSS. Someone even has reported that he used the tbss_non_FA script after using the -T option, and it went fine, by the same error appeared after the -T option. Do you happen to know, did anyone use the -n option and succeeded with a non-FA TBSS?

On Thu, Dec 17, 2015 at 12:32 AM, Daniel Kim <[log in to unmask]<mailto:[log in to unmask]>> wrote:
No need to do preproc step on MD images.

Can you go in your FA folder and try:

$FSLDIR/bin/imglob *_FA.*

You should just get a list of all your subject's FA image names (e.g. "subj1_FA_FA subj2_FA_FA ... subjN_FA_FA")

If you get an error, then something is interfering with imglob.

Danny Kim

On 2015-12-16, at 1:20 PM, Irina Anurova wrote:

Hi Kim,
I have 164 MD images from 164 subjects. All the MD files have exactly the same names as the files in the origdata folder (e.g., SUBJ1_FA.nii.gz).
My subjects are children of different ages, therefore I used the -n option during the FA TBSS (registration to the "most representative" subject's image). Thus the FA directory contains 164x164=26 896 transformation matrices. However, this huge number was not an obstacle for the FA TBSS.
During the FA TBSS we first have to do a preprocessing (by running tbss_1_preproc *.nii.gz). At this step the images are eroded and eventually become smaller in size and are renamed from SUBJ1_FA.nii.gz into SUBJ1_FA_FA.nii.gz. And only after that we do co-registration.
In the user's guide it is not mentioned, should we perform this step (tbss_1_preproc) for the MD images before running the tbss_non_FA script or this erosion and renaming step is included into the tbss_non_FA script?

On Wed, Dec 16, 2015 at 9:36 PM, Daniel Kim <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Hi Irina

How many MD images do you have?
For some reason, imglob failed to create the proper list of input for fslmerge (it complains that the list is too long).

Danny Kim

On 2015-12-16, at 11:08 AM, Rosalia Dacosta Aguayo wrote:

And your FA images, have you checked the registrations?....and what about your MD images?....have you check them too...is you have done this and still have problems...I am sorry to tell you that I do not know how I can help you more unless you are working with two data files for every subject in the space....because the message of fslmerge it is curious to me...please could you explain with more detail what are you really doing? From your information it is not clear for me where can be the problem....

Thank you, and do not deseperate...

Rosala.

2015-12-16 20:00 GMT+01:00 Irina Anurova <[log in to unmask]<mailto:[log in to unmask]>>:
Hi Rosalia,
Yes, I am absolutely sure that my MD images are called exactly the same as my "origdata". The directory MD is created in a correct place. The disk space is not an issue either.
Regards,
Irina

On Wed, Dec 16, 2015 at 5:16 PM, Rosalia Dacosta Aguayo <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Hi Irina,

Are you sure that your MD images are called exactly the same as your "origdata"?

You should get put inside mytbss directory, where you have origdata, stats,....and do: mkdir MD (this will create you a new folder called MD). Once you have the folder, copy inside all your MD images and make sure you have named them exactly as the "origdata".....then you will not have problems when calling tbss_non_FA MD (in order to create your MD template).

Another thing is that you must be careful about spaces, . or something else in the name of your origdata and make an exploration of your FA images in order to make sure they are fine.

Hope this helps,

Rosalia.

2015-12-16 15:44 GMT+01:00 Irina Anurova <[log in to unmask]<mailto:[log in to unmask]>>:
Dear FSL Experts,

When I am trying to run tbss_non_FA on MD images, I receive an error shown below.
I am using the FSL version 5.0. Furthermore, I used the "-n" option for the FA TBSS (tbss_2).
How to solve this problem?

Thank you,
Irina.

The error message:

tbss_non_FA MD
using pre-chosen registration target: A1_F_S_GT065_DTI_raw_FA_FA
upsampling alternative images into standard space
/usr/share/fsl/5.0/bin/tbss_non_FA: 1: /usr/share/fsl/5.0/bin/tbss_non_FA: /usr/share/fsl/5.0/bin/imglob: Argument list too long
merging all upsampled MD images into single 4D image

Usage: fslmerge <-x/y/z/t/a/tr> <output> <file1 file2 .......> [tr value in seconds]
     -t : concatenate images in time
     -x : concatenate images in the x direction
     -y : concatenate images in the y direction
     -z : concatenate images in the z direction
     -a : auto-choose: single slices -> volume, volumes -> 4D (time series)
     -tr : concatenate images in time and set the output image tr to the final option value

Cannot open volume all_MD for reading!
projecting all_MD onto mean FA skeleton
Image Exception : #22 :: ERROR: Could not open image all_MD
terminate called after throwing an instance of 'RBD_COMMON::BaseException'
Aborted (core dumped)