Dear All:
I am analyzing a long scan (240 vols) as individual short epochs (5* 48 vols) using FSL 4.0 FEAT. I
am explicitly modelling the temporal order of the conditions: Epoch 1 (EVs 1 and 2)contains
volumes 1-48, epoch 2 (EVs 2 and 4) volumes 49-97...etc. Each of the 5 epochs contains identical
visual stimulation. EVs in one epoch are only contrasted against each other. The purpose is to see
whether any of the contrasts within individual epochs can show a pattern which is averaged out in
the full scan analysis. To look at this, I have set the highpass filter cut-off spanning the blocks of
the contrasts of interest * 1.5. The number of EVs and contrasts are increased accordingly: Instead
of my 2 EVs, I have 10 EVs. Instead of 4 contrasts, I run 20 contrasts.
I am concerned that there is a temporal drift in the signal that the highpass
filter cannot deal with, which would make epochs from the beginning
incomparable to epochs from the end of the scan. Does anyone have advice on this kind of
analysis ? Many thanks for your input!
Best wishes,
Betina
|