Dear Sante,

Your understanding is almost right.
It doesn't matter if you add up the partial volumes or you take the
mean value over non-zero voxels and multiply by the number of
non-zero voxels - the result is the same.  However, you also
need to multiply by the volume of an individual voxel as the partial
volume values are stored as a proportion of that voxel (e.g. 1.0
means the whole voxel is GM, but is not in units of volume).

So either sum up the partial volume fractions and then multiply
by the volume of a voxel, or multiply the non-zero mean value
by the volume of the non-zero voxels.  Both approaches will
give you what you want, although the latter is a lot easier to
do with fslstats.

All the best,
       Mark

On 12 Dec 2011, at 19:12, Sante Kotturi wrote:

> Dear FSL Experts,
>
> Let me premise this by saying that I am new to this but I have searched the FSL archives and have the following question.
>
> My goal is to calculate the total grey matter volume in an ROI and I have come across two methods:
>
> 1. The first (which is not suggested) is to simply sum values in the ROI in the GM_mod_merg_s3 file.
>
> 2. The second (which is the suggested method) is to multiply the mean GM value in the ROI by the total ROI volume.
>
> It was my understanding that the value in each voxel in the GM_mod_merg_s3 file is the GM volume of that voxel which suggests that simply summing these values would give you the GM volume of all voxels in an ROI. My question is why is the second method highly preferred? Is my understanding incorrect? If so what does the number in each voxel of a GM_mod_merg_s3 file represent?
>
> Many thanks in advance,
> Sante
>
> --
> Sante Kotturi
> University of California, San Diego
> Cognitive Neuroscience
>
>