The idea is that the segmentation of SPM5 warps a set of tissue probability
maps so that they overlay on to the image to segment. After warping, these
will represent (part of) the prior probability of each voxel belonging to a
particular class.
Ideally, these tissue probability maps should represent the prior
probabilities for the population under investigation, and are made by
averaging a large number of spatially normalised tissue class images of
different subjects (wc*.img). I suspect that if you don't have a large
number of subjects, then the tissue probability maps released with SPM5 would
be more representative than an average made up of only about 20 subjects.
Best regards,
-John
> I am doing VBM with SPM5.
> I have created my own customised a priori maps.
> Now, I intend to segment the native T1 images.
> Which of my customised a priori maps should I use?
> 1. the native (unnormalised)a priori maps (c*.img)
> 2. the normalised a priori maps (wc*.img)
> 3. the modulated and normalised a priori maps (mwc*.img)
>
> Because I intend to segment the native unnormalised T1 image
> the native unnormalised segments (c*.img)seem to be the most
> appropriate ones, isn`t it?
|