Widespread correlation of resting state activity after entering more subjects

Submitted by Mamtis13 on Sat, 09/28/2013 - 23:45

Dear DPARSFA and resting state experts,

Having followed your advice on how to set up the pipeline pre-processing for my resting state data (thank you very much for your advice, that was very helpful indeed!), I have now moved on to the fucntional connectivity analysis. Here I am encountering a problem: there is widespread correlation throughout the entire brain with my seed region. However, this result occurs gradually with increasing subject size. I have attached a screenshot of the maps I obtain when entering ever more subjects into 2nd level analysis in SPM.

Is there any chance you might have an idea why I obtain these results? I am not sure how to explain/account for these patterns, your advice would be greatly appreciated!

Kind regards, 
Mamtis

AttachmentSize
Image icon 1_Comparison additive 2nd level rsFC_leftNAcc .png206.61 KB

Submitted by ZHANG_RESTadmin on Wed, 10/02/2013 - 12:37

 your result is quite typical in seed based fc analysis stuides, especially when you selected a small radius for seed region definition.

some processing options will also influence  your result. For example, plain correlation v.s. partial correlation;  not removal global signal v.s. removal of it. .. 

Basically, i would like to say, increase your threshold will make your result more localized. 

Submitted by Mamtis13 on Sun, 10/06/2013 - 20:47

Great, thank you very much for your advice, I will try increasing the threshold.

So far, I did not regress out the global signal given the potential to shift neutral association (i.e. no correlation) between areas into anti-correlation. Would you advise for or against regressing out the global signal as well?

Cheers, Mamtis