Hi,
This is strange because in step 4 you should have copied a matrix
into the reg directory and called it example_func2highres.mat
Therefore it should be able to find this file.
As for where to run the command - it needs to be run in the
feat directory, rather than the reg subdirectory. Part of the
script will cd into the reg subdirectory, and this will fail if it
is in the reg subdirectory already (although it make continue
on and work otherwise, I'd be happier if it ran as expected).
Does it always fail to find example_func2highres.mat when you
run it from the feat directory?
Mark
Julius Fridriksson wrote:
>Mark -- I ran this and it seems to work with one exception: In step 6. I
>get a "could not find example_fun2highres.mat". Isn't it enough to run
>"updatefeatreg . -gifs" under /reg? All updates seem to occur when run
>as such. Thanks for your help on this. Best -- Julius
>
>Julius Fridriksson Ph.D.
>Assistant Professor
>Communication Sciences & Disorders
>University of South Carolina
>Tel: (803) 777-5931
>Fax: (803) 777-3081
>
>
>
>>>>[log in to unmask] 03/10/05 09:53AM >>>
>>>>
>>>>
>Hi,
>
>The best way to do this analysis is still to analyse each session
>individually
>and then combine them later in a higher level analysis. This allows
>both
>the within-session and between-session variances to be modelled
>better.
>However, you are right in that you need to make the registrations work
>specially, as the default in FEAT certainly won't work.
>
>What I recommend is the following:
>1 - pick a reference image (e.g. an anatomical scan from that subject)
>2 - run FEAT on each individual session, but deselect all
>registrations
> except the one to standard space, but use the reference image
>you've
> selected previously as the filename for the "standard space" image
>3 - after FEAT has run, register each session's example_func to the
> reference image separately using FLIRT with cost function
>weighting
> to reduce the effect of the lesions.
>4 - copy the FLIRT matrix from above into the reg sub-directory of the
> appropriate FEAT folder, but call the matrix
>example_func2highres.mat.
> Also create a highres2standard.mat file which contains the
>identity
> transformation - that is the following 4 lines:
> 1 0 0 0
> 0 1 0 0
> 0 0 1 0
> 0 0 0 1
>
>5 - in the reg subdirectory do "imln standard highres"
>6 - do "cd .." (to get to the main .feat directory) and run
> "updatefeatreg . -gifs" here
> This will regenerate the example_func2standard.mat file and
> all appropriate matrices and images along with it.
>7 - Now you can run higher level analyses as normal.
>
>Note that steps 4 and 5 are necessary in this case to trick
>updatefeatreg
>into dealing with just an example_func to standard space registration.
>The key is that if you replace key matrices in the reg subdirectory
>then
>updatefeatreg will fix everything else in this subdirectory to
>incorporate
>them correctly, allowing higher level analysis to proceed correctly.
>
>I hope this makes sense to you, and let me know if you have any
>trouble
>with this.
>
>All the best,
> Mark
>
>
>Julius Fridriksson wrote:
>
>
>
>>Hi -- I have a question regarding co-registration for stroke
>>
>>
>patients:
>
>
>>We employed three pre- and three post-treatment scanning sessions in
>>several persons with large unilateral brain lesions. The plan is to
>>
>>
>run
>
>
>>a higher level analysis for each subject to get mean activation
>>
>>
>across
>
>
>>the three pre- and three post-treatment scans. My quesion is this:
>>should we co-register the three functional runs using FLIRT before we
>>run the actual first level analyses? We would like to mask out the
>>lesion in each case for volume weighting. We are not interested in
>>colapsing across the subjects, only across the three pre- and three
>>post-treatment scanning sessions within each participant. Thanks very
>>much in advance -- Julius
>>
>>Julius Fridriksson Ph.D.
>>Assistant Professor
>>Communication Sciences & Disorders
>>University of South Carolina
>>Tel: (803) 777-5931
>>Fax: (803) 777-3081
>>
>>
>>
>>
|