Hi Steve,
>...As described
>briefly at http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html#tsplot for each
>contrast, at the peak voxel, you get time series plots of all data
>together over all repeats of the event....
Is this exactly true? I ask because it appears to me that there is some sort of thresholding as to
whether an event is included in the tsplot output.
Here's the situation: I have two regressors that are perfectly correlated in time. They are not,
however, correlated in amplitude. One regressor has a constant amplitude, while the other is
modulated from 0 to 1.
When I count the number of trials in ps_tsplot_zstat1_ev1.txt (i.e. the constant amplitude
regressor), I have the correct number. When I count the number of trials in
ps_tsplot_zstat2_ev2.txt (i.e. the modulated regressor), I get something like half the trials that
should be there. Even if I account for those trials that have an amplitude of 0, there are still far
fewer trials than there should be.
This suggests to me that some trials (perhaps those with a small amplitude?) are not included in
the output. Is this correct or am I missing something?
thanks,
jack
|