Hi Mark,
thanks for your help.
Everything works!
I did the registration in the correct way but I applied the final
transformation to the wrong volume (the results of FLIRT on a 4D
volume...).
Now MELODIC is running on my 17 subjects... and I'm waiting for the results!
Thanks again,
elisa
Mark Jenkinson wrote:
> Hi,
>
> You just have to do two separate steps with FLIRT.
> In the first you take two 3D images and determine the registration
> that maps one onto another and save the transformation matrix.
> In the second step to take the matrix from above and use the
> -applyxfm flag to transform your 4D data (-in) to the space of
> the reference (-ref), which will be the MNI152 image if you are
> transforming to standard space.
>
> In the first step you should extract a typical 3D image from your
> 4D image series to act as the input image. If you have done
> motion correction (which is a good idea) then whatever image
> was the reference for the motion correction should be used as
> the input image in the first step (with the reference image for
> the first step being the MNI152 image again).
>
> All the best,
> Mark
>
>
> On 10 Aug 2009, at 15:33, Elisa Molinari wrote:
>
>> I probably found the problem.
>> In registration step I trasformed my 4D data in 3D...
>> Therefore I change my question. How can I apply the FLIRT
>> trasformation to my 4D data saving the correct size? I'm trying to
>> use applyxfm4D... but I've some problem...
>>
>> Thanks
>>
>> Elisa wrote:
>>> Hi all,
>>> this is my first time in your mailing list.
>>> I used FSL to register my fmri data (at rest) to MNI152 and now i'd
>>> like to
>>> make a group analysis with melodic by-passing all pre-processing and
>>> registration steps.
>>> I tried to follow your suggestions (I red in this mailing list) and
>>> using
>>> melodic without GUI specified the file list and concat as approach for
>>> decomposition but it doesn't work:
>>>
>>> melodic -i restfiles.txt -v --report -o groupICA -a concat --tr=0.7
>>> --bgimage=/usr/share/fsl/data/standard/MNI152_T1_2mm_brain.nii.gz
>>>
>>> Melodic Version 3.09
>>>
>>> Melodic results will be in groupICA
>>>
>>> Create mask ... done
>>> Reading data file ../FLIRT/flirted_PREMOT_31_rest_mcf_to_MNI152
>>> ... done
>>> Estimating data smoothness ... done Removing mean image ... done
>>> ERROR:: too many components selected
>>> Do you have any suggestions?
>>>
>>> Thanks a lot,
>>> Elisa
>>>
>>
>>
>> --
>> Elisa Molinari, PhD Robotics Brain & Cognitive Science Dept.
>> Italian Institute of Technology Via Morego, 30 - 16163 Genova, Italy
>>
>> [log in to unmask]
>> phone +39 010 71781479 fax +39 010 7170817
>> www.iit.it
>>
--
Elisa Molinari, PhD
Robotics Brain & Cognitive Science Dept.
Italian Institute of Technology
Via Morego, 30 - 16163 Genova, Italy
[log in to unmask]
phone +39 010 71781479
fax +39 010 7170817
www.iit.it
|