I’m contemplating using the blip_for and blip_rev options to afni_proc. However, before I dive in I’d like to get a better understanding of how to use them properly. From perusing the db_mod.py source code, it seems that each argument needs separate data sets. However, this prompts a couple of questions:
Should both of these two data sets be separate from the data sets provided to the -dsets argument to afni_proc? (I.e., you do not use some of the volumes in your data set of interest (-dsets) as one of the arguments to the -blip_ args, right?
For the blip-up and blip-down files, do yo only need one volume or should more be gathered? If more, is there any consensus or are there any recommendations on the number of volumes that that should be acquired in each direction?
It is great to know you were looking through the code!
The -blip_forward_dset option is not actually needed, as
a corresponding number of volumes would be extracted from
the first -dset dataset if the option is not provided.
The -blip_reverse_dset data certainly needs to be separate,
as it is acquired in the opposite direction of the main,
functional EPI. But if 1 or 10 volumes are provided via
-blip_reverse_dset, then a corresponding 1 or 10 volumes
will be extracted from the first -dset dataset, under the
expectation that the reverse dataset was acquired first,
then the forward datasets.
The basic two options, were are not so strong, are that
you should use either 1 (first pre-steady state) or 10
volumes (from which the median will be used for the
Be careful to acquire the reverse direction data BEFORE
the forward direction data. You do not want to have a
full run of subject motion between the forward and reverse
Thanks for the advice, Rick, much appreciated.
With respect to your pre-steady state comment, I’m using a Siemens scanner which inserts an automatically determined number of dummy scans (distacqs in GE parlance) in proportion to overall scan length, if I recall correctly) before the start of data recording proper. It’s therefore unlikely that much if any pre-steady state EPI data will be acquired. Do you see this as an issue?
Given that, you might opt for the 10-volume method.