Dear Frank,
>Dear John and SPMer's,
>
>one more question about the realigment tool of SPM 99b.
>How is the adjustment option of SPM 96, that take the TR and spin
>history into account, integrated in the SPM 99b realigment algorhythms?
>
>Thank you, regards
>
>Frank
The "motion adjustment" is really a quite pragmatic solution to an
empirical observation that even after correcting for movement (through
reslicing) there is a substantial amount of variance left in the data which
covary with the motion parameters.
When SPM96 was written the current belief was that this residual variance
was due to "spin history" and predicated on this the variance was covaried
out by regressing it onto the movement parameters, and optionally also the
square of the movement parameters.
When SPM99 was written the belief was/is that interpolation artefacts is
the cause of the residual variance, and the strategy for its correction has
been changed accordingly. The new correction regresses the variance onto
functions (sine and cosine) of the displacement of a given voxel center
from the nearest "old" voxel center, which means that a different set of
regressors will be used for each voxel.
Hence, there is no "automatic" way to perform the same motion correction in
SPM99 as in SPM96. However, the realignment module now saves the
realignment parameters in an ASCII file, which you may easily load into SPM
using spm_load. You may then mean correct the resulting matrix and enter it
(and possibly also squared) and optionallly also enter a one-offset version
of it as a covariate of no interest, thereby mimicking quite closely the
correction that was done in spm 96. To see precisely how you should do to
mimic all different options in SPM96 I recommend you looking at the initial
comments in spm_realign.m from the SPM96 distribution set.
Good luck Jesper
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|