Hi,
the nomenclature of the output files follows the conventions of the
"randomise" command. So if you need help on this, have a look there.
You can change the type of statistics (cluster mass, cluster extent,
tfce) inside the dual regression script at the end, where the
randomise command is called.
For me, letting FSL choose the number of components worked best so
far, too. Look for melodic threads in the helpline for more on this. A
typical case of manually setting a number would be in the case of
overfitting, i.e. having equal or more components than time-points in
your measurement, as far as I can see.
Best regards,
Cornelius
On Fri, May 28, 2010 at 5:17 PM, Graeme Schwindt
<[log in to unmask]> wrote:
> Hi Lorena,
>
> I will answer the easiest question -
>
> corrp = corrected p-values - image of p values corrected for multiple
> comparisons (the one you probably want to look at)
>
> I have been letting fsl choose the number of components, simply because I
> don't know how I would choose some maximum, and it has given me a reasonable
> number (~30).
>
>
> I am just about to start on dual regression exploration so can't be much
> help there.
>
> Graeme
>
> On Fri, May 28, 2010 at 10:02 AM, Lorena Jimenez-Castro <[log in to unmask]>
> wrote:
>>
>> Hello FSL experts:
>>
>> Thanks for the previous answer :-)
>>
>> Regarding ICA and dual regression:
>>
>> 1) regarding the output of dual_regression, I would really appreciate very
>> much if somebody could explain to me what exactly means and what are the
>> difference between these images:
>>
>> a: dr_stage3_ic0000_tfce_corrp_tstat1.nii.gz
>> b: dr_stage3_ic0000_tfce_p_tstat1.nii.gz
>> c: dr_stage3_ic0000_tstat1.nii.gz
>>
>>
>> 2) I have been researching the archives but I still have this question:
>> What is better that you choose the number the components or to allow FSL to
>> choose the numbers of them? if the former is better How do you choose a good
>> number, what formula you need to apply?
>>
>> 3) BTW: I know what means "tfce" but What does means "corrp"
>>
>>
>> Thank you very much in advance
>>
>>
>>
>> Lorena Jimenez Castro, MD
>> Postdoc Fellow
>
>
>
> --
> Graeme C. Schwindt, HBSc
> MD/PhD Student
> University of Toronto
> [log in to unmask]
>
>
--
Dr. med. Cornelius J. Werner
Department of Neurology
RWTH Aachen University
Pauwelsstr. 30
52074 Aachen
Germany
Institute of Neuroscience and Medicine
MR Physics - INM4
Research Centre Juelich
52425 Juelich
Germany
::: Please encrypt confidential data :::
|