Dear SPMers,
Working on a set of event related fMRI data i realised that doing the
realign step of the preprocessing was influenced by whether the images to
realign had been "moved" before or not.
Let me explain :
For an unknown reason my EPIs are heavily "shifted" relative to the origin.
So using the "display" tool on SPM 2 i roughly recentered them (e.g x = -100
; y = +100).
It appeared that doing the realign on the original data or on the recentered
ones gives different realignement parameters :
The pattern and magnitude of the movements are the same for all rotation and
translations but that on the Z axis. For the latter the "rencentering"
diminishes the magnitude of movement by more than 50 %!!
By the way the subject being a children, we are talking about BIG movement
(e.g 5 mm), so that's why i think it matters since i may have to include the
movment parameters in my stastistical model!
And besides the pratical reason, i'd like to know why from a theoretical
point of of view this should happen... Realigning just implies matching the
first image of each session to the first image of the reference session and
then matching each image to the first image of the session it belongs to...
So why should it matter if ALL the images have been shifted before?
There are two explanations though :
First being a beginner i've missed on something...
Secondly the difference isn't due to this "recentering" but to something
else...
In any case the answer is the same : i don't what...
Rémi
|