Dear Christian Gaser,
> If you use an absolute threshold > 0.1 most of the spurious clusters outside the brain will
> be removed. These clusters might occur because the probability of being GM is very low
> and these subtle, but sometimes systematical effects can cause effects outside the brain.
>
I guess that already helped.
> Furthermore, the estimated smoothness in your images is much too low (<1mm), which
> points to problems with your model or your data. The estimated smoothness should be
> larger than the applied smoothing filter size.
>
I have no idea what happened here, but I guess I chose the un-smoothed
segmented data for my initial analysis.
When I do it correctly, I get a estimated FWHM of about 15 mm.
Thanks for your help!
Best,
-Lucas
|