Hello,

I've searched the forum and read previous advice on the steps to include two or more contrasts from the same set of subjects - through the "Combine Images" tool. However, I am not certain about a few details when using the new version. Specifically, if I have coordinate locations from two or more contrasts from the same set of subjects (or mostly overlapping subjects), it is my understanding that I should pre-process these contrasts separately (i.e., each having their own text file) and then use the combine images tool with the t-maps that are created.

1. When pre-processing, an upper and lower estimate are created. Should I combine the upper and lower estimates separately (e.g., Study1A_upper + Study1B_upper = Study1_combined_upper, etc.) or combine both upper and lower files from both contrasts?

2. After creating these combined files, am I correct to understand that I need to use the "Convert Images" step and indicate the degrees of freedom?

3. Finally, do I need to re-process this combined images along with the others I am including in the main analysis? Or can I edit my sdm_maps.txt and sdmpsi_params.xml in some way to have it recognize these combined files?

Thanks for all you do to provide this resource to us all!

Best,
Katie

Hi Katie,

I am also interested in combining multiple relevant contrasts from one study (so the same subjects). Did you get an answer to your questions about pre-processing? I have the same questions :) 

Also, is it possible to do this with .txt files (x, y, z, t values) instead of NIFTI images? Or should I be converting my text files to NIFTI images if I want to combine them? 

Thank you! 

Best,
Anna

Hi Anna,

I haven't yet received answers to these pre-processing questions and would very much appreciate hearing from any others if they have successfully done this in the new software version and/or have ideas about what the proper approach is.

Best,
Katie

Hi again,

Has anyone had success with combining two or more contrasts that come from the same set of subjects to be included in meta-analysis using the newest version of SDM? In particular, is it possible to combine them when you only have coordinate locations not t-maps? I cannot find any documentation in the manual or tutorial that applies to this process using the newest PSI version- which creates upper and lower estimates. I appreciate any suggestions others might have.

Best,
Katie

Dear Katherine, Anna,

Unfortunately, we haven't implemented this feature in SDM-PSI yet, I am sorry for this.

I don't know if some of the general recommendations for meta-analyses may be useful for you study, e.g., https://handbook-5-1.cochrane.org/chapte... .

With the best wishes,

Joaquim