Within a single session I collect functional data in five back-to-back
7min runs. I would like to get the best estimate of the haemodynamic
response over those five runs.
In each functional run the TR is 2.5s but with jittering of the onset
of the stimuli we effectively sample the HDR at 0.5s.
If I use Featquery to look at the data from one of the five runs then
I get a nice peri-stimulus plot, with a resolution of 0.5s, that shows
me the shape (and timing) of the HDR. If I use Featquery on a high-
level .gfeat directory that combines the five runs then I get a plot
composed of only a few datapoints which is not very helpful.
There is a similar issue with trying to get an estimate of the lag/
lead of the HDR from the temporal derivatives (ratio of PE1 and PE2
etc). I can get the information from a first-level feat analysis on
one of the five runs, but I can't get an estimate based on the
combined data from the five runs.
I'd prefer not to have to write my own matlab code to solve this
problem, so any suggestions?
(I did try simply concatenating the runs and putting the data into a
single first-level feat analysis. This works sometimes, but comparing
the zstat results to those from an equivalent standard high-level
analysis of the same data, a lot of signal is lost presumably because
there is often too much drift between the runs.)
simon
|