Dear Lisa,
> 1) If I divide my entire scanning time series into multiple scan runs, there
> will be more than 1 GLMs (I suppose multiple sessions, as in lst level
> modelling in SPM, correspond to only multiple GLMs with a mutually
> independent set of onsets and durations for each one.) Thus, for each GLM,
> the implicit baseline will be different from others. So how can I combine
> the contrasts for these GLMs for group level analysis?
The standard SPM approach is to combine all of these scan runs
("sessions", in SPM terms) into a single first-level GLM. The onset
times for a condition refer to the time within this session. So if
you have condition A appear in 3 sessions, you would have 3 columns
for condition A: one column for A in session 1, one column for A in
session 2, one column for A in session 3.
To combine across sessions you simply create a contrast across those
columns (i.e., a 1 over each one of the "condition A" columns).
> 2) The second question is a more practical one. My entire scanning length is
> approximately 40mins. The motive for dividing scanning session into multiple
> runs is to avoid signal decay problem, after scanning for a long time.
> However, what I am not sure is whether it is necessary in my experiment? Has
> anybody used SIEMENS trio 3T scanner under similar circumstance?
I'm not aware of any advantage of multiple (shorter) sessions on
signal strength, nor have I heard about signal problems with 40 minute
sessions. However, there are other practical reasons to break things
up, including the fact that participants can get bored or sleepy, and
the fact that if you encounter a problem it is nice to be able to
salvage some of your data or restart somewhere apart from the
beginning. Though as always, would be interested to hear the
experience of other people here!
Hope this helps!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|