Hi Lino,
I've seen this before with some datasets but it didn't seem that consistent.
It stems from the middle timepoint being treated as the reference under the default metric.
If you switch metrics (e.g., to DVARS) this behaviour should go away.
All the best,
Mark
> On 28 Oct 2016, at 15:43, Lino Becerra <[log in to unmask]> wrote:
>
> Hi:
> I've noticed that fsl_motion_outliers is always detecting the 2 middle volumes as volumes with excessive motion.
> I run it across several subjects and different scanners and get the same error. Inspecting the functional runs (raw or after mcflirt) does not show any significant motion around the middle of the run.
> I compared the output of fsl_motion_outliers from processing done a couple of years ago and have not that issue.
> Wonder if anybody has seen this and what could be the issue. I'm using fsl 5.0.9 on a linux box.
> Thanks!
> Lino
|