Questions about Interpretation of the result(GRF correction and correlation)

Hello, I'm using dpabi well for fMRI analysis. 

Thank you for developing this useful toolbox. 

I have one question about the interpretation of the results in GRF correction. 

Ex) I tried to correlate some clinical outcomes with zfALFF maps in disease group. 

so, I utilized 'correlation anlaysis' function  in 'statistical analysis' tap (zfALFF maps located in 'Group images', clinical outcomes located in 'correlation seed series' for voxel-wise analysis).

Then,  I obtained R.nii and R_cohen_f2.nii files as outcomes. 

And secondly, I tried to perform GRF correction from R.nii in 'Viewer menu'

(I selected mask  and estimated smoothness and selected voxel corrected p<0.01, cluter corrected p<0.05). 

Then, the results found 2 significant clusters(one positive correlation(ex) peak R=0.36), one negative correlation(ex) Peak R=-0.46))

I saved siginificant clusters and mask files(save single cluster one-by-one, 'positive' and 'negative'). 

Finally, I tried to extract mean zfALFF values in each cluster(using mask) among subjects to plot the correlation analysis(X axis=zfALFF in masked area, Y axis=clinical outcomes).

In this case, I applied ROI signal extractor in 'Utilities' menu and I extrated signals in masked areas and correlated with the clinical measures. 

But, the 'R value' in positive correlation  was R=0.45 (ex) and I can't understand the results. 

The peak R was 0.36 in significant postive cluster from R.nii( voxel corrected p<0.01, cluter corrected p<0.05, GRF correction),

so I think the mean positive correlation may be smaller than peak 0.36(at least). 

But the result was higher than peak R(0.45>0.36). 

Also, in case of negative correlation, the result had similar pattern('R'  between extracted mean zfALFF from negative cluster and clinical measure was lower than peak negative 'r' values in negative clusters(ex. -0.57 < -0.46))

So, I'm confused the results and tried to fix the errors that might happen from each step. But I can't find the errors. 

Furthermore, I manually extracted mean signals using matlab code but the results did not change. 

In this case, Is there something I'm missing/error that might happen in these analyses?  How can I interpret this results?

Waiting for the opinions, Sincerely.