See inline responses below.
On Sat, May 30, 2015 at 10:05 PM, Erdem Pulcu Ph.D. <[log in to unmask]>
wrote:
> Dear SPM experts,
>
> - Andy I did not see anyone directly responding to this enquiry, but
> please let me know if you have further information
>
> I am also trying to extract grey matter density information per each from
> the peak of my cluster to plot scatters and linear regression line to show
> the correlation with a performance measure. For functional MRI, I usually
> use Mars bar but I think it does not work with SPM 12, recently Donald
> McLaren suggested peak_nii but I could not make it work with SPM 12 (Full
> release) either, so it all boils down to spm_summarise function now.
>
> I am trying this one below:
>
> [Y, xY]=spm_summarise('beta_0006.nii',struct('def','sphere', 'spec',3,
> 'xyz',[-14 -18 48]'))
>
> I have few related questions:
>
> 1) I am not completely sure whether I should be using the beta or contrast
> or spmT nifti files.
>
Neither, you want to be using the input files to the analysis if you want
to extract the data. The beta files contain the regressor estimates, the
contrast files contain the difference in regressor estimates based on the
contrast vector, and the spmT maps contain the significance of the
contrasts.
>
> 2) this gives me some results in a 1x32 column vector , although my
> analysis contains 29 subjects. How is this possible?
>
You probably have 32 voxels. You asked SPM to provide the beta values for
those 32 voxels as you specified a single image.
>
> 3) the range of these values are between 1.5 and 6.2 (are these in the
> normal range should be expected from VBM-I could not judge and I do not
> want to report something which may be misleading) and what would be the
> unit of measurement? ml/voxel?
>
One can't tell by looking at a beta value whether is in the expected range.
The value depends on the regressor and what you are testing. If you had
extracted grey matter values (from the input images instead of the beta),
then the values should be less than 1, except if you for modulated images
and areas that are mostly grey matter where the values could then exceed 1.
>
> 4) is the ordering of the values in the Y array follow the order of the
> subjects as I entered them to the analysis?
>
The values in Y are the values of each voxel. You only extracted values
from 1 image.
>
> I appreciate peoples comments if anyone is familiar with how to tackle
> this procedure
>
> Many thanks in advance,
>
> Erdem
>
>
>
> ------------------------------
> Date: Tue, 19 May 2015 09:25:10 +0800
> From: [log in to unmask]
> Subject: [SPM] VBM8 Extract GM volume from Peak Voxel
> To: [log in to unmask]
>
>
> Dear all,
>
> I would like to ask how to extract GM volume at peak voxel after doing a
> correlation analysis between 'modulated normalized, non linear only' image
> and behavioral score.
>
> If I clicked 'plot' --> 'contrast estimate and 90% CI', were the numbers
> stored in SPM.y equivalent to the relative regional GM volume for each
> subject?
>
> Best,
> Andy
>
