Hi Shady,
When you use --useScale, the shape analysis will allow the segmented shapes to be 'blown up' (or shrunk) to match the average. This isn't exactly the same as normalising volume, but it is pretty close, so in this scenario local volume decreases are likely to induce apparent increases elsewhere (which may or may not survive thresholding). I think in your case the best thing to do is to drop --useScale, as you have already dealt with head size by using --useReconMNI.
I hope that answers your question and please do let us know if anything is unclear!
Eelke
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
> Of Shady Rahayel
> Sent: 07 December 2016 22:35
> To: [log in to unmask]
> Subject: Re: [FSL] FIRST: abnormal effects?
>
> Thank you Eelke,
>
> I have conducted f tests only to locate the shape-only differences and I get
> many significant results. I use the 'timeseries' method through fslview to see
> the direction of the shape-only differences between my groups: if group 1
> appears to have a positive mean versus the group 2, I take it that group 1
> presents with outward shape deformation for a given peak vertex. I still get
> many instances of concomitant 'outward' and 'inward' shape deformations
> in single structures, many of which are rather surprising yet not impossible. I
> am studying dementia and it seems to me rather unlikely that such patients
> may present with outward deformations while the structure shows clear
> global volume reduction using FIRST. So it brings up a question in my mind:
> once pose and size have been removed from the structure's meshes (using
> both --useScale and --useRigidAlign), are the inward and outward
> deformations still related to volume? In other words once pose and size are
> removed, are we looking at a structure that is well fit in place and with any
> supplementary between-group differences being related to shape (thus
> hypertrophy) and not the structure just being abnormally displaced in one
> direction or the other?
>
> Thank you
>
> Shady
|