Bob,
Thanks for the reply.
We have looked at the data before and after retrend as you suggest.
The data look as expected.
We would gladly send you some .1D files which show time curves before
the fourier filter, after the fourier filter plus retrending and after the fourier filter
with no retrending.
These .1D files would be from voxels that have a non-zero correlation when we do the fim on the retrended data but have a zero correlaton when we do the fim on the non-retrended data.
The key difference is that the non-retrended data are zero centered and thus, many
time points have negative intensities.
I suspect that it is the negative intensities that are causing the problem, although a correlation coefficient should not care if the data are zero centered or not. For some
reason though, the analysis seems to care. At least that is my current theory.
Lee