Hi Javier -
That seems an appropriate approach to me, as long as your selection
procedure for components is not biasing you towards finding the effect
you're interested in in the remaining data.
Cheers,
Eugene
--
Centre for Functional MRI of the Brain (FMRIB) | University of Oxford
John Radcliffe Hospital | Headington
OX3 9DU | Oxford | UK
Ph: +44 (0) 1865 222 523 | Mob: +44 (0) 7946 362 059 | Fax: +44 (0) 1865 222 717
--
On 1 December 2010 10:11, Javier <[log in to unmask]> wrote:
> Dear list,
>
> I would like to check whether the procedure I am following sounds correct. I am analysing some fMRI event-related data, 530 volumes, TR=1.62, 17 subjects. I have done the following:
>
> 1) Run MELODIC to detect artefactual components in the 4D files.
> 2) Make a list of the components to be removed. In most of the subjects I got a total of about 150 components, with about half of them to be deleted.
> 3) After a previous post (032154) replied by S. Smith, I understood that, in the case of having to remove many components, it is better to add them as confound variables in the further FEAT analysis, instead of removing them with the FSL tool; therefore
> 4) I pasted in a txt file the content of the files t2.txt, t5.txt, t9.txt... (for components 2, 5, 9... I found these files in folder melodic/filtered_func_data.ica/report/) and add it as confound variable in the FEAT analysis.
>
> I find 'better' results (they make more sense) after this procedure than removing the components, but I would like to make sure this sounds about right.
>
> As always, thank you very much for your help and your time.
> Warm regards,
>
> Javier
>
>
|