Hi Rick,
I just wanted to follow up with TSNR regarding one of the lines of code for the calculation:
1) 3dTproject -polort 0 -input all.runs.$sub.pb04+orig -censor $out_dir/censor_${sub}_combined_2.1D -cenmode ZERO -ort
$out_dir/X.nocensor.xmat.1D -prefix errts.$sub.tproject.pb04
2) 3dTstat -stdev -prefix rm.noise.all.pb04 errts.$sub.tproject.pb04+orig
On the first line of code the goal is to model the true noise by removing factors we can account for such as motion, censored timepoints etc. In the case above the censored timepoints are replaced with a value zero such that the original length of the timeseries is that same as that of the input signal. Afterwards we calculate the standard deviation and use that in the calculation for the TSNR. My question is if we use cenmode ZERO could this potentially bias the standard deviation results as you are now adding several values of zero to the timeseries which will be calucated into the standard deviation? Would using the option cenmode KILL be more appropriate as we will be removing the timepoints anyways and calculating a scalar value of standard deviation afterwards? Or perhaps cenmode NTRP ?
Thanks,
Ajay