Hi yuanchao,
> Using FIRST package, I can obtain the volume of each
> subcortical structure. I want to make group analysis on
> these volumes controling ICV. So I tried to calculate the
> SCALING factor using sienax command. Then multiply the
> volume of each subcortical structure with the SCALING
> factor. This equals controling the ICV.
Right
> I want to try to calculate the ICV, and fit the
> subcortical volume and ICV into a ANCOVA model to see
> whether it equals controling ICV by multiply a SCALING
> factor. How should I do this?
Unfortunately, there is no easy way to get the intra-cranial volume and you would only get an approximation anyway (see archives on this topic)... You could then just add the scaling factor from sienax as a covariate in the GLM model using the GLM gui and regress it out from your analysis by appropriately setting up the contrasts.
> One additional question:
> By using vertex-wise shape analysis, I can see regions
> of inflation in bilateral thalamus after FDR
> correction, and I detected no volume difference in bilateral
> thalamus. This is not unreasonable,
> since vertex-analysis can localize regional
> abnormalities while volume measure is
> to some extent a whole one and can not detect
> local abnormalities. But using VBM, I
> can detected large cluster of increased gray matter
> densities in bilateral thalamus after FDR corrections. How
> can I interpret the result given no volume difference using
> FIRST vs increased GM density in bilateral thalamus?
How large are these clusters? Because it might be exactly the same explanation than for the shape analysis: VBM might give you very local differences of GM density without the whole volume of the thalamus being significantly increased (though you should maybe see a trend in the volumetry). I think that this is actually quite reassuring that VBM and FIRST shape analysis give you converging results...
Cheers,
Gwenaelle
|