Hi Paul,
I wouldn't recommend this as FIRST is not designed to give a partial volume estimate in the same way that FAST does. This is because the boundaries and contrasts in the subcortical structures are a lot more complex than for the cortex. So we have found it unreliable to use the simple mixture of Gaussian approach that is used in FAST to get good boundaries in the subcortical structures. Instead we rely on information extracted from our manual training set, which does not allow us to derive partial volume estimates (as the manual segmentations never had this).
My recommendation would be to simply use VBM to look for cortical GM-based changes and FIRST vertex analysis to look for structural changes in the subcortical regions.
All the best,
Mark
On 17 Aug 2012, at 07:11, Paul <[log in to unmask]> wrote:
> Dear all FSL experts
>
> I have a question about the possibility of integrating the whole brain GM probability map (the result of FAST) and subcortical segments (the result of FIRST). In current project, i want to use the voxel-wised GM volume as the input features for multivariate classification analysis (using the result after VBM based preprocessing as the input for multivariate analysis).I tried to segment the T1 images from my dataset, i found the subcortical area didn't segment very well using FAST but it can be segmented very well using FIRST. Is it suitable to combine these 2 results into one single NIFTI file for VBM based processing ?
>
> Any command will be very appreciate !!
>
> Best
>
> Paul
>
|