Dear Hamed,
Just to point out it's a little bit more complicated than that. Resel
counts are defined in each dimension of your search space - usually the
last term ("volume") dominates the other ones in the Gaussian kinematic
formula (the result from random field theory providing an analytic form
of the expected Euler characteristic used to assign adjusted p-values to
peaks and clusters) but if your search volume has a very complicated
geometry, the lower terms might get larger (surface area and caliper
diameter) and penalise your inference.
See the equation at the bottom right of Keith's poster:
http://www.math.mcgill.ca/keith/poster7.jpg
Best regards,
Guillaume.
On 02/12/16 15:20, hamed nili wrote:
> Dear all,
>
> thanks for the replies.
> Here is my recap of the last few points:
>
> 1. As most of you said, the FWER can be affected by this and it need not
> go down:
> According to Gaussian RFT, FWER is proportional to the number of voxels
> in the volume divided by the product of the smoothness in the 3
> directions (i.e. x, y and z).
> What I am doing is reducing V (the numerator), but also changing the
> denominator.
> So ... the effect could go either way. Furthermore using a GM mask can
> be motivated in cases where we are not interested in performing
> hypothesis testing in what I call irrelevant voxels anyways.
>
> Also as Bruno pointed out this would not affect the non-parametric tests
> (e.g. using SnPM) and perhaps also FDR.
>
> 2. I agree with the last point that activations in irrelevant regions
> might be a hint towards poor pre-processing or GLM in some subjects at
> least.
>
> 3. I totally agree with the point about smoothing.
>
> Bests
> Hamed
>
>
>
>
>
> On Fri, Dec 2, 2016 at 2:37 PM, MRI More <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
>
> Dear Hamed,
>
> In case your functional volumes were smoothed I would smooth the GM
> files accordingly before applying a threshold to generate the mask
> (or alternatively, mask the functional volumes and only then apply
> smoothing). The decision should be based on where GM is present in
> the processed images, not where GM was present initially.
>
> I wouldn't go with any such masks though (at least as long as you
> don't have acquired functional data with very high resolution). The
> unmasked maps provide some useful information. "Activations" in
> ventricles and/or WM might point to some problems with drift or head
> motion. Once you apply a mask only the boundary areas of the
> artefact might remain and look like "nice" activations in basal ganglia.
>
> Best regards
>
> Helmut
>
>
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|