Hi there,

I should do one easy task - compare correlation before and after rTMS treatment between two selected ROIs (restfMRI). I used CONN to extract Z-values  for each subject
and I also wanted to try to extract values from ROIs by myself using SPM functions. The preprocessing was done using CONN and I extracted values separately from wau* and *swau images. I used mean values from each ROI for each time point and get pearson correlation coeficient between ROIs and then the atanh transformation was done to obtain Z-value of correlation. ... In CONN firstlevel analysis the Z-value for two selected ROIs was -0.4745 but in my extracted ROIs it was 0.0393 for wau images and 0.0272 for swau images.

What could be the reason for such huge difference? Does CONN pipeline do something different to compute correlation?

Thanks for any clarrification,
Matyas

Why are you using smoothed data for a ROI-to-ROI analysis?  You should only use the unsmoothed data in these cases, from what I understand.

Jeff

I tried both smoothed (swau) and unsmoothed (wau) images to try, which ones conn uses. I understand that unsmoothed data are better for such analysis. My question is why is there huge difference in correlation computed by CONN (z values in first level analysis) and correlation computed by myself. If there is something I am doing wrong when computing correlations (as described in my first question) ....

Hi again,

now I understand that I was comparing correlation from not preprocessed files (wau* files) and correlations that were computed after denoising by CONN.

Nevertheless, I compared data extracted from ROIs from results/preprocessing/niftiDATA_Subject*_Condition000.nii and data saved in results/preprocessing/ROI_Subject*_Condition000.mat - and values are different. What is the difference between values saved in DATA_Subject* and ROI_Subject* files?

Thanks for reply,
Matyas