I have a question about how L1 analysis is done in AFNI. The afni_proc script seems to integrate preprocessing and L1 analysis of the data so that after this script, the data are ready for group analysis. For a pipeline that our team work on, we are wondering if it is possible to separate the preprocessing and L1 analysis parts in afni_proc, so that preprocessing is done in a different pipeline, but L1 analysis is conducted in AFNI? If so, we need to understand 1) which parts in afni_proc are for L1 analysis (3dDeconcolve?), and 2) whether or not AFNI L1 analysis is compatible with nifti file format instead of the head and brick format?
Thank you in advance!
Sure, you can specify the processing blocks that you want included with the -blocks option. But how to do this reasonably depends on exactly what you want included in the regression model. And for censoring, it would be important to include a 6 column motion parameter file (-regress_motion_file). But if you apply all of the same -regression options that you would use if afni_proc.py were doing all of the processing, plus the motion file, it should be mostly there.
Note that there are some aspects that might be worth paying attention to, such as if any tshift operation might have be done (and what effect that might have on the stimulus times), global or voxelwise scaling, any extra regressors that should be included, etc.
If you post an afni_proc.py command, that would also provide more to comment on. Plus, it might help to describe the actual preprocessing steps that will already be run.