[MINC-users] Estimating tissue volume from PVEs

[Andrew Janke]

Hi Andrew,

2008/9/3 Andrew Reid <A.T.Reid at science.ru.nl>:
> I want to get an estimate of tissue volume from the PVE volumes in my
> minc directories. As far as I can tell, this requires that I:
>
> 1. Mask with the brain mask (using Perl):
> $mask = $subject_dir."/".$sname."/mask/".$prefix.$sname."_brain_mask.mnc";
> `mincmath -mult ${volume} ${mask} ${temp_vol}`;
>
> 2. Resample with the inverse of the linear talairach transform:
> $transform =
> $subject_dir."/".$sname."/transforms/linear/".$prefix.$sname."_t1_tal.xfm";
> `mincresample -transformation ${transform} -tfm_input_sampling
> -invert_transformation ${temp_vol} ${temp_trans}`;
>
> 3. Calculate tissue volumes. For this I bin the volume into ranges of
> size 0.1, and multiply the resulting sum by the mid-point of the bin,
> adding two bins for [0:0.01] = 0 and [0.99:1] = 1. Then I sum the
> resulting values.
>
> My questions are:
> 1. Is this the correct way to apply the inverse transform?

Yes, as far as I can see.

> I get a weird
> result in that the mean native GM volume actually decreases with respect
> to that obtained from the original volumes in talairach space, but the
> mean native WM and CSF volumes increase. This is confusing, given that
> it is a linear transform, which I believe should deform in a globally
> proportional way.

That is correct, it should be proportional. perhaps there is something
else askew further down the track in the volume calculations.

> 2. Is there a better (more continuous) way to calculate volume than my
> step #3? The binning method essentially smooths the data, instead of
> simply summing the PVE values from all voxels.

Hrm, If I can grok what you are doing correctly why not just this?

minccalc -expression "(A[0] < 0.01) ? 0 : ((A[0] > 0.99) ? 1 : A[0])"
${temp_trans} ${temp_trans}.clamped
mincstats -sum ${temp_trans}.clamped

At which stage you will have to multiply the result of the mincstats
call by the volume of a single voxel and possibly include a mask if
you want the volume of a particular chunk and/or your ${temp_trans} is
not a single structure already.



-- 
Andrew Janke - andrew.janke at anu.edu.au
Department of Geriatric Medicine, ANU
(a.janke at gmail.com || http://a.janke.googlepages.com/)
Canberra->Australia    +61 (402) 700 883