Dear Dr. Yan,
I am using DPARSF-A to generate fALFF data from the raw-resting state data. You already suggested remove nuisance covariates as early as possible, i.e. before fALFF calculation.
Now I wonder about advantages of smoothing, normalizing and detrending my data before (compared to after) fALFF-computations. DPARSF-A allows for both options. In recent papers, I can find both ways. Which way do you suggest, and why would you advice to use one way or the other?
Thank you a lot for your advice and for clarifying pros and cons of the two approaches,
Kati
a network for supporting resting-state fMRI related studies.
Hi Kati,
Hi Kati,
I will prefer to put Smoothing and Detrending before fALFF-computation.
Smoothing could increase signal-to-noise ratio. Detrending is needed as linear trend will affect FFT or filtering process.
Normalizing can be put before or after. Usually I put normalizing before smoothing as it's a pretty convenient option.
Best,
Chao-Gan