Hi Kaitlyn,
I am not sure that I understand your design correctly - is it the case that E1 and E2 are your EVs, and that each of these contains 4 trials of the same type?
In any case, this does sound a lot like a motion-related problem. My suggestion would be to use fsl_motion_outliers (see http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSLMotionOutliers) to remove the effect of the volumes that are badly affected by motion to see if that makes the ‘activation’ disappear.
Cheers,
Eelke
> On 9 Jul 2015, at 19:38, Kaitlyn Breiner <[log in to unmask]> wrote:
>
> Hi FSL users and experts,
>
> I'm finding too much activation for one contrast in one subject. I ran a first-level analysis to assess differences between two time points (E1 and E2). The poststats report revealed virtually no activation for E1 v Baseline (not too surprising), but the complete opposite for E2 v Baseline across the whole brain. E2 v E1 is worse, with nearly the entire brain showing activation. A few things are worth noting:
>
> 1) There are 4 trials that are modeled for each event.
> 2) This exact contrast was used for 26 other subjects, and twice more for different contrasts for this subject and did not reveal this problem.
> 3) Registration is good.
> 4) The onset files look fine, and after running the first-level analysis on this contrast again, the same issue arose.
>
> There appears to be some motion that occurs around the time points of these events, but I'm doubtful whether that's the cause. Any help/advice would be greatly appreciated!
>
> Kaitlyn
|