De-mean? | The R-fMRI Network

Submitted by linjh on Sat, 05/10/2014 - 17:38

Dear sir(s),

I want to use the data released at NITRC as my database (normal control) to further my study about resting network.

However, some bias I should correct.

As I have known, a strategy like de-mean might solve my problem.

The detailing might be like (take an example like: I would like to check clevealnd case<cle1>'s resting network in comparison to leipzig--lpz1,2,3...)

I should calculate lpz1-lpz1(mean), lpz2-lpz2(mean), lpz3-lpz3(mean).......lpzn-lpzn(mean) -----> De-mean

I should sum these resuts and calculate the Mean (de-mean lpz), Std (de-mean lpz)

Then I cold check the Z of Cle1 = (Cle1-Cle1.mean)- Mean(de-mean lpz)/ Std (de-mean lpz)

------

Am I right? But I have met a problem. How can I get the lpz1(mean), lpz2(mean)....lpzn(mean) as well as Cle1...

Is any one can resolve my problem?

Thank you so much!

Or Could I use mReHo?

Dear sir(s),

I have read some "unofficial" document that said mReHo = ReHo/mean of whole brain (mean).

Is it true? (I have seen the description "mReHo is the raw ReHo value divided by the global mean value.
Thus, most mReHo value could be near "1". If you compare to "0", then you can find high value everywhere. which mentioned by 嚴老師.

How is "the the global mean value" calculated?
!

-----------------------

Then, May I use the process like :

mReHo = ReHo/mean --> mean = ReHo/mReHo

--> For a single case, ReHo-mean= ReHo-ReHo/mReHo = ReHo *(1-1/mReHo) = de-mean of a case

--> Then I can get the mean of the sum of the de-mean in Leipzig (LPZ) group (Mean_de_mean) and the stanadard deviation of the de-mean (Std_de_mean)

--> Then I can use the LPZ data to verify my data (for example case1: Z= <(ReHo1-ReHo1/mReHo1) - Mean_de_mean>/Std_de_mean) to correct the differences between LPZ's setup and mine?

Thank you so much!

Hi,

For comparing across different sites, use mReHo or zReHo are acceptable stadardization methods, however, site effects remain. You may need to model the site effects in the group analysis still. Here is a paper discussing standardization across sites: Yan, C.G., Craddock, R.C., Zuo, X.N., Zang, Y.F., Milham, M.P., 2013. Standardizing the intrinsic brain: towards robust measurement of inter-individual variation in 1000 functional connectomes. Neuroimage 80, 246-262.

Best,

Chao-Gan