If you're using echo time to identify artifactual components, you will want to check out this paper:
"Differentiating BOLD and non-BOLD signals in fMRI time series using multi-echo EPI."
http://www.ncbi.nlm.nih.gov/pubmed/22209809
On Friday, November 09, 2012 05:11:31 you wrote:
> Hi Benjamin,
>
> thanks for the reply and suggestion.
> I think you are right, simply splitting up the mixing matrix for each subject is doesn't work.
> I'll try dual regression to get the individual maps.
>
> may I ask another follow-up/extended question?
> we used multi-echo sequence to acquire the functional data.
> we assume that the first echo (16ms) mainly contains noises while the second (36ms) signals.
> (and we did see so in results of ICA based on one single subject data).
> thus we want to regress out the noise components found in TE1 from TE2.
> if we use group ICA to find out the noise components in TE1, how can we then use dual regression (or other ways) to backproject the group (noise) components from TE1 to individual subjects (TE2)?
>
> thanks a lot!
>
> Best,
> zzang
>
|