Hi,

I'm not sure whether you want to correlate the changes that are happening with time within a session or make some summary measure of changes for the whole session and then use this summary measure for a higher level analysis.  One difficultly with such designs is if the changes are very slow then they get confounded with slow drifts in the signal and can often be removed by the temporal filtering if you are not careful.  However, it depends on the nature of your stimuli.  If it is simply watching a video (or a series of clips without any breaks between them) then you will have slowly varying signals as I describe here and it will be quite difficult to separate these from drifts and you will lose a lot of power in temporal filtering.  Two ways to avoid this are to either alternate video clips with fixation or some other task, so that you have things changing more often and the size of these changes can be used, since they are not subject to the slow drifts.  Another way would be to use ASL (or some other acquisition variant that included a way of calibrating/correcting for slow drifts).

If your question is simply about how to include a measure in the higher-level analysis, then it is simply a question of calculating the value and including these values, after demeaning them, in an EV at the higher level (in addition to a group mean EV).  However, getting the right first-level analysis might still be tricky, depending on what your stimulus is like.

I hope this is helpful.
If you still have some questions then feel free to send another email but include some more details about your design and what you are trying to achieve.

All the best,
	Mark
 

On 18 Sep 2014, at 19:19, Mayte Parada <[log in to unmask]> wrote:

> Hello FSL experts!
> 
> I'm fairly new with FSL and I'm interested an efficient way of correlating body temperatures recorded during my scans with activation data. Temperature tends to increase when participants watch arousing video clips and not so much when watching other types of videos. The tricky part is that temperature doesn't go back to baseline once the person becomes aroused during the arousing clip. It will drop steadily over time (and throughout) the other clips and will then increase again when the next arousing video is played. Its not as simple as just hypothesizing that temperature is greater during the arousing clips!
> 
> If anyone encounters this kind of issue with other types of arousal that they have some feedback on, please share. So far one idea is to take the area under the curve, compare to baseline temps. How could I input this into FSL in a higher level analysis?
> 
>