help
help > RE: Contrasts selection for timeseries extraction
Feb 25, 2021 01:02 PM | kitty zhu
RE: Contrasts selection for timeseries extraction
Hi,Dr. McLaren
Thanks for your reply. It makes me understand I mixed up some definitions before, and now I can describe my question more clearly.
I mixed the contrast which is used to define ROI up with the one used to adjust for time-series extraction when I use spm_region to generate VOI*.mat file.
I want to define individual ROIs which is bounded within the group-level ROI,here is the part in the script :
%extract VOI
%----------------------------------------------------------------------
matlabbatch{1}.spm.util.voi.spmmat = fullfile(outputDir,'SPM.mat');
matlabbatch{1}.spm.util.voi.adjust = 1;
matlabbatch{1}.spm.util.voi.session = 1;
matlabbatch{1}.spm.util.voi.name = 'rIFGtri6mm'; %'rIFGtri' 'lIFGtri' 'lbilateral' 'rbilateral'
%thresholded SPM
matlabbatch{1}.spm.util.voi.roi{1}.spm.spmmat = {''};
matlabbatch{1}.spm.util.voi.roi{1}.spm.contrast = 4;
matlabbatch{1}.spm.util.voi.roi{1}.spm.threshdesc = 'none';
matlabbatch{1}.spm.util.voi.roi{1}.spm.thresh = 0.005; %0.001
matlabbatch{1}.spm.util.voi.roi{1}.spm.extent = 5;
matlabbatch{1}.spm.util.voi.roi{1}.spm.mask = struct('contrast', {}, 'thresh', {}, 'mtype', {});
%group level sphere mask
matlabbatch{1}.spm.util.voi.roi{2}.sphere.centre = [48 35 8]; %Atypical [48 35 8];typicalIFG[-42 26 26] bilateral
matlabbatch{1}.spm.util.voi.roi{2}.sphere.radius = 10; %15
matlabbatch{1}.spm.util.voi.roi{2}.sphere.move.fixed = 1;
%individual level that was constrained by group mask
matlabbatch{1}.spm.util.voi.roi{3}.sphere.centre = [0 0 0];%not used
matlabbatch{1}.spm.util.voi.roi{3}.sphere.radius = 6; %6mm
matlabbatch{1}.spm.util.voi.roi{3}.sphere.move.local.spm = 1; %move to nearest local maxima in group mask
matlabbatch{1}.spm.util.voi.roi{3}.sphere.move.local.mask = 'i2';
matlabbatch{1}.spm.util.voi.expression = 'i1&i3';
The spm_region conducts VOI definition and time_series extraction, so the output file VOI*.mat include two information.
I want to know whether I understand the analysis process correctly:
First I use spm_region to get VOI*.mat file ,and then in gPPI script ,
P.VOI only read VOI information in the VOI*.mat, and following timeseries extraction is then conducted using
P.extract='eig'
P.contrast = 'Omnibus F-test for PPI Analyses'.
Is that right?
Thanks in advance!
Kitty
Originally posted by Donald McLaren:
Thanks for your reply. It makes me understand I mixed up some definitions before, and now I can describe my question more clearly.
I mixed the contrast which is used to define ROI up with the one used to adjust for time-series extraction when I use spm_region to generate VOI*.mat file.
I want to define individual ROIs which is bounded within the group-level ROI,here is the part in the script :
%extract VOI
%----------------------------------------------------------------------
matlabbatch{1}.spm.util.voi.spmmat = fullfile(outputDir,'SPM.mat');
matlabbatch{1}.spm.util.voi.adjust = 1;
matlabbatch{1}.spm.util.voi.session = 1;
matlabbatch{1}.spm.util.voi.name = 'rIFGtri6mm'; %'rIFGtri' 'lIFGtri' 'lbilateral' 'rbilateral'
%thresholded SPM
matlabbatch{1}.spm.util.voi.roi{1}.spm.spmmat = {''};
matlabbatch{1}.spm.util.voi.roi{1}.spm.contrast = 4;
matlabbatch{1}.spm.util.voi.roi{1}.spm.threshdesc = 'none';
matlabbatch{1}.spm.util.voi.roi{1}.spm.thresh = 0.005; %0.001
matlabbatch{1}.spm.util.voi.roi{1}.spm.extent = 5;
matlabbatch{1}.spm.util.voi.roi{1}.spm.mask = struct('contrast', {}, 'thresh', {}, 'mtype', {});
%group level sphere mask
matlabbatch{1}.spm.util.voi.roi{2}.sphere.centre = [48 35 8]; %Atypical [48 35 8];typicalIFG[-42 26 26] bilateral
matlabbatch{1}.spm.util.voi.roi{2}.sphere.radius = 10; %15
matlabbatch{1}.spm.util.voi.roi{2}.sphere.move.fixed = 1;
%individual level that was constrained by group mask
matlabbatch{1}.spm.util.voi.roi{3}.sphere.centre = [0 0 0];%not used
matlabbatch{1}.spm.util.voi.roi{3}.sphere.radius = 6; %6mm
matlabbatch{1}.spm.util.voi.roi{3}.sphere.move.local.spm = 1; %move to nearest local maxima in group mask
matlabbatch{1}.spm.util.voi.roi{3}.sphere.move.local.mask = 'i2';
matlabbatch{1}.spm.util.voi.expression = 'i1&i3';
The spm_region conducts VOI definition and time_series extraction, so the output file VOI*.mat include two information.
I want to know whether I understand the analysis process correctly:
First I use spm_region to get VOI*.mat file ,and then in gPPI script ,
P.VOI only read VOI information in the VOI*.mat, and following timeseries extraction is then conducted using
P.extract='eig'
P.contrast = 'Omnibus F-test for PPI Analyses'.
Is that right?
Thanks in advance!
Kitty
Originally posted by Donald McLaren:
Hi
Kitty,
In any PPI, you are extracting the timeseries for the entire functional run. The difference between different extractions is often how the timeseries is adjusted. In gPPI, it is recommended to only adjust for variables of no interest (e.g. motion parameters). To do this, one uses the omnibus F-contrast. P.Contrast='Omnibus F-test for PPI Analyses'. This is the default behavior.
The gPPI scripts will extract the timeseries and then generate the gPPI model for you. There is no need to extract the timeseries separately.
Best,
Donald
Originally posted by kitty zhu:
In any PPI, you are extracting the timeseries for the entire functional run. The difference between different extractions is often how the timeseries is adjusted. In gPPI, it is recommended to only adjust for variables of no interest (e.g. motion parameters). To do this, one uses the omnibus F-contrast. P.Contrast='Omnibus F-test for PPI Analyses'. This is the default behavior.
The gPPI scripts will extract the timeseries and then generate the gPPI model for you. There is no need to extract the timeseries separately.
Best,
Donald
Originally posted by kitty zhu:
Dear gPPI experts,
I have some questions about the contrast selections for timeseries extraction before running the gPPI script.
In the sPPI tutorial(the attention example), they use the motion contrast (include attention and non-attention conditions) to extract timeseries for following attention-modulated connectivity analysis. In gPPI, all conditions and corresponding PPI items are modeled in the GLM, so I wondered whether I need to extract timeseries including all conditions in gPPI GLM model . For example, in above attention example, using function spm_region, I should extract timeseries from F contrast(1 0 0 ;0 1 0;0 0 1) incuding all conditions rather than just attention effect(0 -1 1) contrast, is that right?
kind regards,
Kitty
I have some questions about the contrast selections for timeseries extraction before running the gPPI script.
In the sPPI tutorial(the attention example), they use the motion contrast (include attention and non-attention conditions) to extract timeseries for following attention-modulated connectivity analysis. In gPPI, all conditions and corresponding PPI items are modeled in the GLM, so I wondered whether I need to extract timeseries including all conditions in gPPI GLM model . For example, in above attention example, using function spm_region, I should extract timeseries from F contrast(1 0 0 ;0 1 0;0 0 1) incuding all conditions rather than just attention effect(0 -1 1) contrast, is that right?
kind regards,
Kitty
Threaded View
| Title | Author | Date |
|---|---|---|
| kitty zhu | Feb 18, 2021 | |
| Donald McLaren | Feb 22, 2021 | |
| kitty zhu | Feb 25, 2021 | |