Mark -- Thanks very much for your help. Sincerely -- 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] 03/11/05 02:06PM >>>
OK, it is a difference in the updatefeatreg script.
Sorry about that.  Yours will only work in the reg
subdirectory as it is written and my local one prefers
the feat directory but will probably work in both.
It appears that the version I have here has been updated
and isn't the same as our distributed version.

So you should use yours in the reg subdirectory and
forget what I said about going to the feat directory.
Once again, apologies for any inconvenience.

All the best,
        Mark


On 11 Mar 2005, at 18:37, Julius Fridriksson wrote:

> The two commands below do not work:  dirnm=`pwd`; > command not
found;
> cd ${dirnm}/reg > undefined variable. I am attaching a copy of my
> updatefeatreg, anyway. Any suggestions? Thanks -- 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] 03/11/05 01:03PM >>>
> Hi,
>
> The script is written so that the variable dirnm is set to the
> argument.
> It then tries to do "cd ${dirnm}/reg" internally.
> For this reason the argument should be the feat directory.
>
> It is true that I could have written it to run from the reg
> subdirectory,
> but I didn't.  I tried to keep it more consistent with other feat
> scripts
> that run from the top level feat directory.
>
> So I'm slightly concerned that it does not work from the feat
> directory
> but does work from the reg subdirectory.  This doesn't really make
> sense and makes me wonder if it is an indication of some other
> problem which may cause headaches in the future.  For this reason
> I'm reluctant to just say run it from the reg subdirectory and don't
> worry about the problem.  I'd rather track down the cause of the
> problem
> now in case it affects something else later.
>
> It is also true that it probably will work from the reg
subdirectory,
> as
> the "cd ${dirnm}/reg" will fail and you'll just stay in the current
> directory, which is fine.  But I can't see why it fails to work from
> the
> feat directory.  Maybe it is because `pwd` is returning something
> strange.  Try the following commands at the prompt, from the
> feat directory:
>   dirnm=`pwd`;
>   cd ${dirnm}/reg
> Does this get you to the reg subdirectory?
>
> If not then that is the problem.  If is does work though, then I'm
> still quite confused as to why the script isn't working as it
should.
>
> So, could you please run the lines above, and also send me a copy
> of $FSLDIR/bin/updatefeatreg so that I can check that it is the
> same version as the one here?
>
> Thanks,
>     Mark
>
>
>
> Julius Fridriksson wrote:
>
>> It script does not seem to continue into /reg when run from the
feat
>> dir. It fails everytime I run this from the feat dir. Why is it
>> necessary to run this from the feat dir rather from /reg? Thanks --
>> 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] 03/11/05 12:37PM >>>
>>>>>
>>>>>
>> 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
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
> <updatefeatreg>