AFNI version info (
afni -ver): Version AFNI_23.0.04
I have a question about the APQC. I am trying to isolate the QC info (e.g. motion, SNR, etc) for just the timepoints for my stimuli that I use in my 3ddeconvolve regression. I have found the motion.[subj].enorm.1d file, but am hoping to get summary statistics (like the qsumm APQC output) for just the time points I am interested in. Is that included somehwere in the output and I am just missing it? Or is there a feasible way to generate that information?
Thank you so much for your help!
The APQC HTML itself is meant to be quite general, so it isn't feasible to tweak the SNR to the subset of "during stimulus" intervals.
But note that in the "warns" QC block, one of the checked items to report+warn about is the fraction of "during stimulus time points" that have been censored due to motion. So, that is something stimulus-interval specfic (and might highlight potential stimulus-correlated motion, for example, if not just decreased statisticality for the stimuli).
You could mirror the statistics and outputs that are calculated for the subset of time points in which you are interested. The degree of calculation for each of those varies per item. The TSNR volume, for example, is calculated by afni_proc.py's proc.* script itself, so you can see how that is done and add the extra information of which timepoints to select with AFNI subbrick selectors, for example:
The qsumm section at the very end of the report has per stimulus class statistics such as "num TRs per stim" both (orig) and censored, as well as "fraction TRs censored" and "ave mot per sresp".
Note that those apply to stimulus RESPONSE curves, not stimulus event timing, specifically.
This is really helpful! Thank you both!
Actually, Rick, I don't see anything about censoring TRs per stim or fraction TFs censored in the section on per stimulus class statistics. Only num trs per stim and ave mot per sresp (see screen shot).
I can't quite figure out what I am missing...
Oh, it looks like no censoring options were provided. Were you planning on censoring high-motion time points? If so, please paste (or message me with) your afni_proc.py command.
The average motion (0.198) is moderately high, so censoring would seem reasonable. There is basically no difference in motion between the 2 stim classes though.
Rick, you are right. It looks like I accidentally left out the regress_censor_motion flag from my proc.py script. Thank you for catching that and for all of your help today!