Hi!
Your questions are very general and the best answers will depend on the
context.
1) The frequency peak is usually lower than 0.05 Hz for the resting state
networks including the DMN. In many DMN paper spectra of such RSNs are
printed. The exact shape of the fft spectrum depends on how you
preprocessed the data including the temporal filtering.
2) To get a feeling of how the resting state data are transformed by ICA
and to see, how many components they 'need' to be represented, you should
run a number of single subject ICAs with restriction of the numbers.
3) Too general question as many methods are now ICA based but then have
different facets. Consider group ICA (time concatenaed) mode if you
analyse a whole group. Best also see the FSL research pages for MELODIC.
Hope this helps as starter,
Philipp
|