normal ditribution before RFX ananlysis

Submitted by Dinos2010 on Mon, 10/04/2010 - 13:52

Hi all,
In the article "Detection of functional networks in the resting brain" there was a step before group analysis involving transforming the individual Reho data to normal distribution. Has this function been implemented in REST ? If yes, what specific method is being used?

Thanks,
Hua

Submitted by zangzx on Thu, 10/07/2010 - 13:43

Dear Hua,
Recently, we have done a study (unpublished) on the distribution of ReHo cross participants. We chose 100 participants and performed the ReHo on REST. We then performed the Kolmogorov-Smirnov goodness-of-fit hypothesis test and found that the ReHo results of over 97% voxels in mask are normaly distributed. So we think that the ReHo results could be used in group level statistic analysis without transformation.
I hope it can help.

Zang Zhen-Xiang

Submitted by Dinos2010 on Thu, 10/07/2010 - 22:14

Hi,
Thank you so much for your reply. Another question : the method of bandpass is ideal rectangular filter but there are many other way of filtering, such as fir, hamming...
which is the optimal for rest fmri signal?

cheers,
Hua

Submitted by YAN Chao-Gan on Fri, 10/08/2010 - 16:03

'which is the optimal for rest fmri signal? '
Then a definition of 'optimal' need to be defined.
To my knowledge, there is no study giving the goldern standard for filtering methods. Maybe you can perform a study to evaluate which way is the best.

Submitted by ZangYF on Thu, 10/14/2010 - 07:12

There are many options for any of the many parameters in fMRI data preprocessing. Some of them might be critical to the final results. However, there has been no good criteria for "what is optimal".