The low pass filter is not required, just set the lpf cutoff freq to 0. For detrending, set the nuisance regressors with the -xreg flag. The polynomial detrending order is set by the highpass cutoff freqency. You can play with it to get a 5th order polynomial. There is no equivalent to the -hpf in makanalysis, but the highpass filter in fcseedcor should come close. There is not a tpexclude option, but you could create a regressor file with the time point of interest set to 1 and all other tps set to 0. This might get you closer than trying to grow something at home.
On 6/15/17 7:54 PM, Lee, Erik Gordon wrote:
Hi All,
I am trying to do some seed to seed functional connectivity calculations that replicate my analysis design in fsfast. I know FreeSurfer 6 has the fcseedcor function, but it appears that this requires a low pass filter, which is not a part of our processing stream. Also I don't think the fcseedcor function has any detrending which is a part of our analysis.
I am putting together a Matlab script to do this for me, and want to be sure that the steps are equivalent.
This is my normal mkanalysis-sess command:
mkanalysis-sess -analysis targ.2cm.rh -surface fsaverage rh -fwhm 6 -notask -taskreg target_2cm.dat 1 -nuisreg vcsf.dat 5 -nuisreg wm.dat 5 -mcextreg -polyfit 5 -fsd rest -TR 3 -per-run -hpf .01 -tpexclude tpe_file -nuisreg global.waveform.dat 1
This is what I am doing in Matlab:
-First, I apply a polynomial detrending to all the seeds and nuisance regressors (first five components from vcsf and wm, first three from mcextreg, and one component of global signal), while ignoring any bad timepoints. -Second, I make all signals evenly sampled again by applying a linear interpolation over missing timepoints for both the seeds and nuisance regressors. -Third, I apply a high order high pass filter. -Fourth, I get rid of all the timepoints that were generated from the linear interpolation (now post filtering). -Fifth, I use a glm to regress all the nuisance variables from the ROI signals. -Sixth compute the correlation coefficients between seeds using the residuals from the previous step.
How similar is that to what is done in fsfast? Any tips on what to change?
Thanks! Erik
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer