Hi everyone,

I have a general question on meta analysis on neuroimaging studies. If my interest is in differentiation within one region, can we input coordinates of just one region across studies? Can SDM do it with a mask of this region? Just like ROI?


Thanks in advance.


Qian

Dear Qian, SDM limits its meta-analytic calculations to the voxels within a mask. Default SDM masks are generic (e.g. gray matter), but you can use your own mask. To this end, preprocessing must be called as a command (e.g. using the bottom command line input) rather than pressing the button.

First copy your mask to the share directory, and change its name so that it begins with "mask_" (for example "mask_myROI.nii.gz"). Afterwards, go to SDM and type the pp command:

pp correlation_template, degree_of_anisotropy, isotropic_fwhm, mask, number_of_randomizations

where correlation_template may be e.g. "gray_matter" (or any other "cor_" file from the share directory), degree_of_anisotropy may be 1, isotropic_fwhm may be 20, and mask may be the name of your mask. For example:

pp gray_matter, 1, 20, myROI, 20

Hi,

Thanks for the reply and information.

I am interested in task based activation, rather than structural studies such as VBM or DTI. Can I still do the analysis on one region using SDM?


Best,
Qian

Yes, of course  ;-)

Hello! I have a similar question. I am doing a functional MRI meta analysis and would like to do a ROI analysis, ignoring other areas. I looked through the tutorial and the manual but could not find the description on how to perform this analysis. Do I specify ROI analysis starting from preprocessing? If so, how to do it?Suggestions are much appreciated. 
Thank you
Peggy

Dear Peggy,

you should first create a standard, binary mask, name it "mask_roi.nii.gz", and copy it to the sdm share directory. Afterwards, start the pre-processing selecting this mask.

Best,

Joaquim

Thank you very much! -Peggy

Thank you for your comment. I wonder if this method is applied to analyze just a single portion of the brain, would that justify a higher alpha level than the default p=0.005? Or would using a smaller mask already 'automatically' relax the significance level an appropriate amount?