Dear Paivi
With a TR of 2s I think the best you can do is to specify the onsets in seconds (note: this does not need to be an integer value), then convolution should take care of the rest. Your real problem would have to do with the fact that different slices are acquired at different timepoints within the TR. This could to some extend be solved with a slice specific design matrix but this is currently not implemented in SPM, and introduces a few other problems.
Best
Torben
> Den 18. feb. 2020 kl. 19.32 skrev Paivi Eerola <[log in to unmask]>:
>
> Thank you. Well, now I understand what microtime is doing, and it is doing it to the regressors, not the fMRI signal.
>
> In my case, I have a set of participants listening to musical pieces in randomized order. Once the scanning session starts, participants listen to several musical pieces (with a minute of silence between them) without the scanning session stopping or without synchronizing the initial fMRI pulse to the onset of each piece (note: the pieces are at least 1.5 minutes long).
>
> I want to perform time series correlation amongst subjects fMRI signals for each musical piece.
>
> Question: how to get the fMRI time points related to the respective onset times of the start of the musical pieces? In other words: how could I extract the fMRI signal for each participant and piece, so they are all aligned?
>
> The problem here is precisely that musical piece starts on a different time for each subject, so the onset can be anytime within the TR of two seconds.
>
> Solutions: Could we perform a microtime resolution but on the fMRI signal? This is, could we interpolate time points in the fMRI time series as to find the exact moment where the musical piece starts for each participant. Does this make sense?
>
>
> Thank you,
> PE
|