Dear Christian
Thanks a lot for your reply. I try to clarify my last email.
The picture attached was a "melodic_IC.nii.gz" file which we registered
according to your command line suggestion. We used the
MNI152_T1_2mm_brain template (the same used in the Melodic/Registration
tab) and the example_func2standard.mat as transformation file:
flirt -in melodic_IC.nii.gz -out test.nii.gz -ref /usr/local/fsl/data/standard/
MNI152_T1_2mm_brain.nii.gz -applyxfm -init example_func2standard.mat
As you could see, the IC-map did not fit very well onto the template. Is it
possibly normal that the fit isn't perfect and the images don't overlap exactly?
In another try, we split the melodic_IC.nii.gz file to obtain a single volume as
input. We did this because in our further analysis we need the single volumes
of each subject. The single volumes resulting from this split were named with
the initials of the subjects, e.g. AC***.nii.gz. This worked, but the following
registration (same parameters as above) did fail with the following error
message:
** ERROR (nifti_image_read): failed to find header file for 'AC001'
** ERROR: nifti_image_open(AC001): bad header info
Error: failed to open file AC001
ERROR: Could not open image AC001
Image Exception : #22 :: Failed to read volume AC001.nii.gz
terminate called after throwing an instance of
'RBD_COMMON::BaseException'
(we checked for spaces in path and filenames and there were none)
The exact sequence of operation was:
We conducted single session ICA on every subject with the default settings in
the melodic GUI (except TR = 2.5 s, high pass cutoff = 50 s and spatial
smoothing FWHM = 3 mm), i.e. the settings in the registration tab were:
# Registration to standard image?
set fmri(regstandard_yn) 1
# Standard image
set fmri(regstandard) "/usr/local/fsl/data/standard/MNI152_T1_2mm_brain"
# Search space for registration to standard space
# 0 : No search
# 90 : Normal search
# 180 : Full search
set fmri(regstandard_search) 90
# Degrees of Freedom for registration to standard space
set fmri(regstandard_dof) 12
# Do nonlinear registration from structural to standard space?
set fmri(regstandard_nonlinear_yn) 0
# Resampling resolution
set fmri(regstandard_res) 4
We hope we expressed ourselves more cleary. Do you see any problem with
our way of proceeding?
Cheers,
Carmen
On Fri, 20 Mar 2009 21:44:44 +0000, Christian F. Beckmann
<[log in to unmask]> wrote:
>Dear Carmen,
>
>Difficult to say what is going on without more specific information.
>The picture you send is the 'output' of what exactly? Are you trying
>to overlay this on the MNI152 template brain at 2mm isotropic
>resolution?
>
>You clearly did not use fslspit on melodic_IC.nii.gz, the error
>message says that fslsplit cannot find AC001 - are there maybe spaces
>in your full path?
>
>Also, what is the exact sequence of operation? Here is what I think
>you should be doing:
>
>(i) either (a) use separate melodic runs on the individual data sets -
>when using the GUI the resulting .ica directories should contain a /
>reg sub-directory containg all the necessary transformation file which
>you can apply to the IC maps as described in the previous email prior
>to feeding them up to a higher level analysis
>
>OR
>
>(b) use a single melodic GUI set up with 'time series concat' to
>perform all pre-processing/registration and resampling steps
>automatically. Even if you're not interested in the final concat
>output the run should give you all functional runs transformed into a
>common space. You can then run separate ICAs on these filterd and
>transformed data sets
>
>(ii) concatenate all the maps you want into a single 4D file.
>
>(iii) run melodic on this - no need for any further processing or
>transformations.
>
>hth
>Christian
>
>
>
>
>
>On 20 Mar 2009, at 14:59, Carmen Ghisleni wrote:
>
>> Dear List
>>
>> thanks for the answer, we tried this approach but unfortunately it
>> didn't work
>> out as we hoped.
>>
>>
>> In a first attempt, we tried to use an image of a single component
>> found by
>> the first ICA. This image resulted from using fslsplit on the
>> melodic_IC.nii.gz.
>> This didn't work out at all and we got the following error:
>>
>> ** ERROR (nifti_image_read): failed to find header file for 'AC001'
>> ** ERROR: nifti_image_open(AC001): bad header info
>> Error: failed to open file AC001
>> ERROR: Could not open image AC001
>> Image Exception : #22 :: Failed to read volume AC001.nii.gz
>> terminate called after throwing an instance of
>> 'RBD_COMMON::BaseException'
>>
>>
>> In a second attempt, we used the original, "unsplit"
>> melodic_IC.nii.gz-file.
>> Here we didn't get an error but the output wasn't satisfactory, I
>> attached a
>> picture of the result we got.
>>
>>
>> Maybe here lies the catch: Is the melodic_IC.nii.gz-file the correct
>> file to use
>> and if not which file should we choose as input?
>>
>>
>> Cheers,
>> Carmen
>> <Bild1.jpg>
>
>_______________________________________________
>Christian F. Beckmann, DPhil
>Senior Lecturer, Clinical Neuroscience Department
>Division of Neuroscience and Mental Health
>Imperial College London, Hammersmith Campus
>Rm 419, Burlington Danes Bldg, Du Cane Road, London W12 0NN, UK
>Tel.: +44 (0)20 7594 6685 --- Fax: +44 (0)20 7594 6548
>Email: [log in to unmask]
>http://www.imperial.ac.uk/medicine/people/c.beckmann/
>
>Senior Research Fellow, FMRIB Centre
>University of Oxford
>JR Hospital - Oxford OX3 9DU
>Tel.: +44 (0)1865 222551 --- Fax: +44 (0)1865 222717
>Email: [log in to unmask]
>http://www.fmrib.ox.ac.uk/~beckmann
|