Hi,
I have a question related to exploratory ICA-based denoising for single subject data.
I know there's an option in FEAT for exploratory ica, but for some reasons I did not use that option at the first place.
after running feat, I aligned the subject image with MNI standard image and then run melodic for exploratory ica.
The component numbers from gui-based ica and post-feat ica are different.
my question is,
1. if it is okay to run feat first, normalize the data with standard template and then to do exploratory ica?
2. at single subject level, the number of components from exploratory ica sometimes is as many as up to 150 or so, which requires much effort to differentiate noise from signal. Would it be okay to specify dimensions in ica for the purpose of denoising?
thanks for the help.
Best,
Wei
|