Correlating a quantitative variable with WM estimates

AFNI Message Board
1

Hello AFNI team,

I am working with a white matter data set of 10 subjects. We are looking to find the correlation between white matter values and one quantitative variable (news event scores) across participants for each voxel. Each individual has their own score but they are not assigned into groups, therefore it would be a correlation across all subjects.

I am wondering what the appropriate group level modeling command would be since there is no between-subjects factor? Currently, I was thinking it would be either 3dLME or 3dMVM with the following commands below. Is it possible to use these commands on 3D data and not 3D+time data? If not, what would the recommended program be to complete this voxel-wise correlation?

Thank you,
Catherine

3dMVM
3dMVM -prefix Example
-bsVars ‘1+NE’
-mask myMask+tlrc
-dataTable
Subj NE InputFile
s1 54 s1+tlrc’[WhiteMatter]’
s2 60 s2+tlrc’[WhiteMatter]’
…
s10 70 s10+tlrc’[WhiteMatter]’

3dLME
3dLME -prefix Example
-mask myMask+tlrc
-model “1+NE”
-ranEff ‘~1’
-dataTable
Subj NE InputFile
s1 54 s1+tlrc’[WhiteMatter]’
s2 60 s2+tlrc’[WhiteMatter]’
…
s10 70 s10+tlrc’[WhiteMatter]’

2

Catherine,

If I understand your situation correctly, you have a simple regression model

WhiteMatter = a + b * NE + e

If this is true, 3dttest++ would be good enough for this.

3

Great, thank you Gang!