Let it estimate them, 150 is fine if you have plenty of data. Keep in
mind that ICA+FIX will only clean up spatially specific noise. Global
physiological noise is not visible to spatial ICA, and would have to be
cleaned using some other means.
Peace,
Matt.
On 2/20/17, 1:09 PM, "FSL - FMRIB's Software Library on behalf of Marzia
Restuccia" <[log in to unmask] on behalf of
[log in to unmask]> wrote:
>Dear FSL experts,
>
>I am a new FSL user, and I am performing a study on task-based, ultrafast
>fMRI data (TR = 400 ms). I'd like to use the MELODIC ICA + FIX tool to
>identify physiological noise components. I have several datasets from
>different subjects, I performed ICA on one subject and the algorithm
>identified more than 150 components. Do you suggest to limit the number
>of components (if yes, how many components should I set?) or to let
>Melodic to automatically estimate it?
>
>Thank you in advance.
>
>Best regards,
>Marzia
|