Hi Prof Yan,
I am an undergraduate student in Singapore. I have asked you some questions before and I would like to express sincere thanks for your helpful replies.
At current stage, I am trying to perform the following data processing on HCP data for functional connectivity analysis. I tried several tools (FreeSurfer, SPM, REST, DAPBI) and did not find an intuitive one. Now I tried RESTplus, which seems to be the latest updated one, and consider it to be suitable for my work.
Different from normal batch pre-processing, my data is retrieved from HCP and is already partially pre-processed with a so-called 'minimum pipeline' and is already in MNI coordinate. Also, I prefer to use Matlab functions instead of clicking on GUI button.
Could you give me some suggestions on how to perform the following Pipelines? Since the documentations and comments for those codes are missing.
1. Detrending and demeaning across each run
Is the function function rp_detrend (ADataDir, APostfix , CUTNUMBER) the correct one for detrending? or spm_detrend? How about demeaning? Are they the same?
2. Bandpass filtering(0.01 - 0.08Hz)
Is the following function the correct one to use?
function [] = rp_bandpass(ADataDir, ...
ASamplePeriod, ALowPass_HighCutoff, AHighPass_LowCutoff, ...
AAddMeanBack, ...
AMaskFilename,...
CUTNUMBER)
3. Head-motion regression and whole-brain signal regression
Is the function function [] = rp_RegressOutCovariates(ADataDir,ACovariablesDef,APostfix,AMaskFilename) the correct one to use?
4. 6-mm FWHM smoothing
Can you tell me what function to use for smoothing? Is it the function function spm_smooth(P,Q,s,dtype) in SPM?
Please tell me if there is more efficient way using Freesurfer or other tools for my tasks.
Thank you for your time. I would really appreciate for your help.
Best Regards,
Hongquan Long
Hi Hongquan,
Hi Hongquan,
You are correct for the functions. Alternatively, if you are using DPABI, then it's:
1. Detrending and demeaning across each run:
y_detrend. However, you can put it into nuisance regression, as a polynomial trend.
2. Bandpass filtering(0.01 - 0.08Hz)
y_bandpass
3. Head-motion regression and whole-brain signal regression
y_RegressOutImgCovariates
4. 6-mm FWHM smoothing
You can use spm_smooth. You can also setup one step smooth in DPARSF.
Best,
Chao-Gan
Hi Prof Yan,
Hi Prof Yan,
Thank you for your reply. However, I have another question regarding to y_bandpass.
This function has the following parameter:
% AAddMeanBack 'Yes' or 'No'. if yes, then add the mean back after filtering
% Last Revised by YAN Chao-Gan, 100420. Fixed a bug in calculating the frequency band. And now will not remove the linear trend in bandpass filter (as fourier_filter.c in AFNI), but just save the mean and can add the mean back after filtering.
Detrend usually means
Detrend usually means removing the linear trend.
Here in y_bandpass, the mean will be removed (but not the linear trend).
Add mean back doesn't change anything in the later calculation, as we are interested in the fluctuations rather than the mean. You could add it back if you want to visuallize the processed images.
Best,
Chao-Gan