Dear SPM experts,
I have a question regarding PSC transformation in a design with HRF + time derivative.
I was oriented towards Cyril Pernet's very helpful scripts (see here: http://www.sbirc.ed.ac.uk/cyril/bold_percentage/BOLD_percentage.html).
If I understand correctly, for event-related designs with HRF + time derivative, Cyril Pernet relies on a standard trial (one 1-sec single event as a 'default currency') to derive the max event height for the PSC-scaling factor, while for a blocked design, one should take the actual duration as defined in the design matrix.
I have a mixed blocked-event-related design, and would like to get maximum comparability between block- and trial-related effects (although I'm aware of the fact that they shouldn't be compared directly).
So in order to compute the percent signal change values for event-related effects, I would like to derive the scaling factor for the events with their actual duration as well (instead of using the standard-trial-duration).
In my design matrix, one single event should have a mean duration of 2.463sec (mean RT for this event).
Is the following procedure correct?
1) request the basis functions with time resolution adapted to my design matrix (dt = 32/16 = 0.0563)
clear all
xBF.dt = 0.0563; % time resolution adapted to the design matrix
xBF.T = 32; % adapted
xBF.T0 = 16; % adapted
xBF.name = 'hrf with time derivative';
xBF.Volterra = 1;
xBF.length = 32;
xBF.order = 1;
xBF = spm_get_bf(xBF)
2) go with the standard-trial procedure to derive the 'combined_event'
combined_standard_event = xBF.bf * ones(1,size(xBF.bf,2))'; % use all basis functions
3) then convolve the event from my design matrix, i.e., with the actual duration = 2.463/ dt (--> thus in my case /0.0563 = 44 sample points), with this combined vector:
Reference_trial = conv(ones(1,44),combined_standard_event)';
Any help would be highly appreciated.
Thank you very much in advance!
Best,
Sandra
|