Hi SPMers,
I am running into a problem that I am trying to figure out with difficulties and thought I could ask. I am running an fMRI task and I am interested in the AMG response to both negative images vs baseline and negative images vs neutral images.
I have extracted information about this brain structure in 3 ways:
_______________________________
1) Using Marsbar 1:
_______________________________
D = mardo(spmfolder);
R = maroi(roi_name);
Y = get_marsy(R, D, 'mean');
xCon = get_contrasts(D);
E = estimate(D, Y);
E = set_contrasts(E, xCon);
b = betas(E);
marsS = compute_contrasts(E, 1:length(xCon));
[e_specs, e_names] = event_specs(E);
n_events = size(e_specs, 2);
ets = event_types_named(E);
n_event_types = length(ets);
bin_size = tr(E);
fir_length = 24;
bin_no = fir_length / bin_size;
dur = 0;
opts = struct('single', 1, 'percent', 1);
n_events = size(e_specs, 2);
pct_ev = [];
% Return percent signal esimate for all events all subjects in design
for e_s = 1:n_events
pct_ev(e_s) = event_signal(E, e_specs(:,e_s), dur);
end
[ets, names] = event_types_named(E);
pcs_data(:,i) = pct_ev;
________________________
2) Using Marsbar:
________________________
Pulling the SPM design and getting the value of the contrasts directly from the second level.
_________________________
3) Using SPM:
_________________________
Getting the eigenvariates for the region of interest for the contrasts of interest
I would expect the 3 values to be very highly correlated, but they are not.
I am sure I am not the first one to run into this issue and it has been discussed elsewhere, but I am having difficulties finding it (there's a broken threat here https://sourceforge.net/p/marsbar/mailman/message/23164015/ that doesn't seem to continue :( )
I will very much appreciate any thoughts and insights into what sort of information each method provides and what leads to this lack of correlation between them.
Thanks tons in advance!
Vicky
|