I hope you don't mind me cc-ing this to the list, but it might be relevant
for others too, and it would be very helpful to get some more feedback on
I did get a few responses to my query, below, some saying it is not really
necessary to add regressors for the 'day' effect, and some saying you do
have to model all possible contributors to variance (as was my own p.o.v. as
What I ended up doing was to model the different scan days (consisting of 4
runs each) as sessions and then adding 4 regressors to each session, one for
each run). I have attached the resulting estimated model. I am very
interested to hear if people think this is a good way to go for modeling
longitudinal data for single subjects, with multiple runs per scan day.
My design is slightly different from the one you describe in you email
(below), as I have my 3 conditions present in each run, and you seem to have
different conditions in different runs. The latter is a separate issue in
itself, but w.r.t. the modeling I don't think it makes a difference.
%%From: Jérôme Redouté [mailto:[log in to unmask]]
%%Sent: Thursday, June 05, 2008 4:03 AM
%%To: Dirk den Ouden
%%Subject: Re: [SPM] longitudinal model
%%I read your question on the SPM mailing list, and as I did
%%not found any response can I ask you how you finally managed
%%your model specification for a single subject longitudinal study?
%%Did you enter your day scans as Runs and then modelized your
%%conditions within your actual runs as conditions?
%%for ex, if I had 3 scan days . Each day , 4 runs with first 2
%%runs: cond A vs. Rest (block design) and 2 last runs: cond B vs. rest.
%%Would you enter data in SPM as follows:
%%3 sessions and 4 conditions in each sessions (A, RestA, B, RestB)?
So, I would say, 3 sessions, with 3 conditions (A, B, Rest) each, and 4
extra regressors (one for each run) per session.
%%Thanks for your help
%%Dirk den Ouden a écrit :
%%> Dear all,
%%> For a longitudinal fMRI study in single subjects, I want to
%%> data of two or three different scan days in one model, so as to be
%%> able to compare main effects and contrasts on different
%%days (in single subjects).
%%> There are 4 sessions (runs) within each scan day, and SPM of course
%%> adds the session-specific effects in the model.
%%> 1) Is it possible to model the day-specific effects as well
%%> (basically, to add a separate column for each scan day in
%%> or in other words, to add sessions over sessions)?
%%> - or would you advise to just treat the 4 sessions
%%within one scan
%%> day as one session, and model the different scan days as if
%%> different sessions?
%%> 2)Do you agree that my model should have day-specific regressors?
%%> Thanks for all help!
%%> Dirk den Ouden
%%> Dirk-Bart den Ouden, Ph.D.
%%> Aphasia & Neurolinguistics Research Laboratory
%%> Dept. of Communication Sciences & Disorders
%%> Northwestern University
%%> 2240 Campus Drive, Evanston IL 60208-3066
%%> Phone: (office) 1-847-467-2515; (lab) 1-847-467-7591
%%CERMEP - Imagerie du vivant
%%Centre d'étude et de recherche multimodal et pluridisciplinaire
%%59 Bd Pinel
%%69667 Bron - FRANCE
%%tel : 33 (0)4 72 68 86 13 (bureau)
%%tel : 33 (0)4 72 68 86 00 (standard)
%%fax : 33 (0)4 72 68 86 10