I don’t usually post brief updates like this, but a recent update to afni_proc.py has me really excited. As of the May 13th 2014 binaries of AFNI, you can now expect your resting state data to process considerably faster thanks to a new-ish program called 3dTproject.  3dTproject is meant to replace 3dDeconvolve for resting state processing and shows huge improvements in speed!  How fast do you ask?  What normally takes my computer (12-core Mac Pro) about an hour to process for resting state was accomplished in a mere 10 seconds using 3dTproject.

How can you get the improvements?  Update your AFNI binaries (@update.afni.binaries -d) and use afni_proc.py for your processing.  Alternatively you can call 3dTproject by hand on your data.  You’ll need to run 3dDeconvolve with with the -x1D_stop, which will setup the necessary matrices (i.e.. X.nocensor.xmat.1D) for 3dTproject to use.  Once this is done, it’s as simple as giving 3dTproject your input files, the Xmat and a censor file and you’re off to the races!

3dTproject -polort 0 -input pb04.$subj.r*.blur+tlrc.HEAD \
 -censor censor_${subj}_combined_2.1D -cenmode ZERO \
 -ort X.nocensor.xmat.1D -prefix errts.${subj}.tproject

That’s all for now!