Hi Shady,
This is definitely something that can happen and as you mention, it most likely represents an overall displacement of the structure. When using --useReconMNI, FIRST relies on the affine registration to the MNI template for alignment. This means that using this mode, differences in the position and pose of a structure can be detected in addition to differences in shape.
If you want to investigate this further, you can use the --useRigidAlign option, which will do another registration step using just the segmented shapes. This will remove any differences in position and pose.
Hope that helps,
Eelke
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
> Of Shady Rahayel
> Sent: 23 November 2016 06:17
> To: [log in to unmask]
> Subject: [FSL] FIRST: abnormal effects?
>
> Hi,
>
> I have recently conducted several analyses using FSL-FIRST. I am questioning
> myself however about the between-group results I tend to have with
> subcortical structures. Indeed, in many structures, results appear as
> opposite-sign clusters. For instance, I can get one cluster whose scalar values
> may be negative as well as an opposite cluster, not always symmetrical but
> almost, showing positive scalar values. Is this a spurious effect that has been
> observed in FSL-FIRST analyses before such as consequence of some
> abnormalities in preprocessing? Does it necessarily suggest problems with
> the data? Am I wrong to interpret it as a combination of shape displacement
> and expansion? I did check every step of preprocessing and haven't found
> any 'abnormalities'.
>
> I always used vertex analysis with --useReconMNI option.
>
> Thank you very much,
>
> Shady
|