Dear Wil,
> I would appreciate any and all comments on the following line of
> thinking about image realignment in fMRI time series data.
>
> If there is scanner-based signal drift in a time series, and lets
> assume it is linear, then the signal distribution in the final image
> will be, necessarily, different from the first image (i.e., reference
> image). If we assume a case of no displacement, the realignment
> algorithm will minimize the difference between first and last images
> treating signal-drift as displacement (i.e., head movement). So, would
> it not be more appropriate to remove any signal-drift before image
> realignment is performed?
I think this line of thinking is very appealing. However there may be
a danger of removing movement-related drifts that are the basis of the
movement estimation. We will think about this more deeply.
With best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|