Dear Ce,
> I am trying to estimate the signal intensity of individual trials in a
> rapid ER design. Intuitively, I would think of establishing the
> 1st-level GLM, using one regressor for one trial.
> For example, let say I have 100 trials in one run/session, then when
> defining my 1st-level GLM, I would add 100 conditions along with 100
> onsets respectively for these 100 trials. However, the concern is that
> there might be too many regressors. ( In fact, if I introduce temperal
> derivative into the model to account for hemodynamic delay, I would have
> 200 regressors altogether.) Is this a problem? Or has anybody tried to
> fullfill similar purpose ( estimating single trial signal intensity
> maps) before?
Yes, this is definitely possible. How effective this is probably depends
on the details of your design - for example, how many total scans you
have. If you have 100 events and 100 scans (which would not be a good
design anyway...), this will not work very well. If you have 100 events
and 10,000 scans, you are in no danger of having too many regressors.
Probably your case lies somewhere in the middle. You are correct that
you may want to refrain from including the temporal derivative to save
the degrees of freedom.
Also keep in mind that for any single event, the estimates of activity
you get will be very noisy for one subject, no matter how many scans you
have.
Hope this helps!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Center for Cognitive Neuroscience and
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|