Hi Geoff,
From a quick glance at SPM2's spm_spm.m I think the beta images
written to your output directory are the Weighted Least Squares
(non-sphericity corrected) versions, so I think if you read the
equivalent voxel from these, with e.g.
V = spm_vol(spm_get(inf,'.img','pick beta images'));
voxel = [x y z 1]'; % with your (x,y,z) coords
y = spm_get_data(V, voxel);
They should match the plotted ones. I hope...
Best,
Ged.
> I am using SPM2 and need to extract the betas from each subject in a
> second level analysis to analyse in SPSS. In an analysis without non-
> sphericity correction it is easy – I go to my voxel, plot the effects of
> interest and then type "Y" in the matlab window. This gives me a long
> column which is number of subjects by number of regressors long (in my
> case 24 subjects by 4 regressors = 96). I then put these numbers into
> SPSS.
>
> When I use non-sphericity correction I do the same thing but the numbers I
> get out do not match the plot I get - i.e. the mean of the first twenty
> four numbers does not match the value for the mean plotted in the graph
> for the first regressor. I'm guessing that this has something to do with
> the correlation matrix but as my maths is appalling that's as far as I got.
|