Dear Alessandro,
Volkmar will correct me if I'm wrong but, yes, the option to
replicate/scale across sessions is an all or none. In your case, the
"T-contrast (cond/sess based)" could be an option as you can specify
which sessions to replicate over but, if all of your sessions have the
same form, it might be easier to specify the contrasts over sessions by
hand:
sess = [1 1 0 1 0 0 1 0 ...]; % whether to include session or not
cond = [1 -1 1 -1 1 -1 1 -1]/4; % within-session contrast
c = kron(sess, cond) / nnz(sess); % scaled contrast over sessions
If you also included covariates (such as movement parameters) or used
anything beyond the canonical HRF, you would also have to take these
into accounts - this is what the "T-contrast (cond/sess based)" would do
for you automatically.
Best regards,
Guillaume.
On 23/06/2021 09:48, Alessandro Zanini wrote:
> Hi SPMers,
> I'm new to this list and I'm sorry if someone already asked this
> question, but I cannot find the info on the net. I'm performing
> 1st-level fMRI analyses in SPM12 for a within-subject study. In my
> design, each participant performed 10 runs per experimental condition,
> with 3 different conditions globally. Then, I come out with 30 sessions
> per participants. I used an adaptation design, and in each condition I
> have 8 possible combinations of stimuli to contrast.
> At the moment of the contrasts, I face the problem of the replication
> over sessions: is it possible to replicate & scale the contrast X, but
> only over N sessions? I would to perform the same contrast over the 10
> sessions of the first condition, scaling them, but I can only choose the
> option "don't replicate" or "replicate and scale", and I cannot specify
> any session interval. This means that I only can replicate the same
> contrast either over the 30 sessions or nothing? In this second case,
> does the absence of scaling be a problem in my analysis?
> I also found the option to perform T contrasts based on
> sessions/condition, and I think that it can be a good choice, but I
> found almost 0 info and tutorials about it..
> Thank you in advance!
> /Alessandro Zanini/
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
London WC1N 3BG
|