What I would like to do is deconvolve the time series for each condition, so that I can take a seed region from each individual time series and then do a simple correlation analysis within that time series.
In other words, deconvovle condition A, B, C, and D from the original time series, so that I now have four separate 3d time series, one for each condition.
Then, for example condition A. I would use the time series of a seed voxel in the deconvolved time series for condition A to do simple correlational analysis within condition A.
Does that make sense?
Lesley