Dear all,

I just tried the example 9 of the latest afni_proc.py (as of yesterday), and was very happy to see the addition of the segmentation step (lending some hope for a semiautomatic, freesurfer-less ANATICOR?). I noticed, though, that the tissue classes appear to be mislabeled in the segmentation output datasets, as follows:

GM (1) -- is really --> CSF
WM (2) -- is really --> GM
CSF (3) -- is really --> WM

On a different topic, I also noticed that for a fairly standard EPI resting state acquision (one 200-volume run, TR=2 sec), the bandpass step of the script models the filter with 125 (!) sin/cos functions, on top of the motion parameters and their derivatives, yielding over 140 regressors (for 200 volumes). Now, that seems to me a bit excessive, so I wonder if, unless you have several catenated runs or a very long one, it may be really best to skip bandpassing.

thanks for your thoughts!

best
giuseppe

giuseppe