According to your email, it seems that the maximum difference in acquisition time between slices is very short. Therefore, it seems that slice time correction will not make a huge difference.
In order to optimize the first-level model specification, you can set up the microtime resolution and onset according to the number of slices. Set up the microtime resolution to the number of slices you had if the entire TR interval would be acquiring slices, i.e. TR / (time between slices). Then, set up the microtime onset to half the number of slices.
For instance, let's say that your time between slices is 100ms but you have acquired only 10 slices during the first 1s.
Then, set up your microtime resolution to 30 (=3/0.1) and microtime onset to 5.
Hope this helps,
Cesar Caballero
On Mar 1, 2011, at 8:22 AM, Jonas Persson wrote:
> Dear SPM community,
>
> We're currently using an EPI-sequence for our fMRI study, with a TR=3. However, it's set up to acquire all the slices as fast as possible during the beginning of TR and then "relax" for the remainder of the time (as opposed to equidistant acquisitions, where the slices are evenly distributed throughout the TR). We considered this to have the advantage of giving less problems with time distanses between slices, but could this pose a problem with regards to the way SPM handles the data, i.e. if SPM assumes equidistant acquisition? Also, could this be problematic for event-related designs?
> We don't plan on using any slice time correction.
>
> best regards,
>
> Jonas