Nuissance covariates in Alff/fAlff

Submitted by aguinea on Sat, 02/16/2013 - 00:33

Hi,

I want to make an Alff/fAlff analysis using the last version of DPARSFA with images previously pre-processed (realigned, coregistered, normalized and smoothed), but not using this software. Moreover, I want to covariate the analysis with some physiologic data, like cardiac and respiratory signals. In DPARSFA seems to be the chance of introducing the nuisance covariates before, but also after, making the Alff/fAlff analysis. The introducing these  nuisance covariates before or after Alff/fAlff box is going to modify the Alff/fAlff output images??

On the ther hand, I don't have very clear if the filter option is going to affect these nuisance covariates. 

Thanks,

Andrés

Submitted by conge on Sun, 02/17/2013 - 01:07

 Hi Andrés,

I am not sure if anyone every tried to compare the ALFF/fALFF results of regressing out phisilogical signals before and after ALFF/fALFF calculation. You might want to try that by yourself to find out. In all the papers I've read so far about ALFF/fALFF, however, nuisance regression usually happens before filtering and f/ALFF calculation.
Bandpass fitering resting state fMRI data is believed to be a way to remove physilogical noise. In my mind, fitering should happen after nuisance regression, although different people might have different opnions on this.

Cheers,

Yang

Submitted by YAN Chao-Gan on Wed, 02/20/2013 - 10:40

Hi Andrés,

The nuisance covariates regression after ALFF/fALFF calculation will never affect the ALFF/fALFF images.

In one recent ongoing work, we do observed regressing out head motion helps in reducing the relationship between ALFF and motion.

Best,

Chao-Gan