Questions about EPI variance line warnings

Dear AFNI message board:
In the process of data processing, I found that EPI variance line warnings showed no line found. Could you please tell me the meaning of "no line found"?
thank you very much!!


As part of this large Demonstrating QC in FMRI project, we looked in depth at things to check during FMRI processing. One of those that we saw a number of times were odd lines of high variance typically going through the acquired plane in a narrow frequency+phase region through the slice planes. These have strong effects on cross-brain correlation patterns, which were discussed here:

  • Reynolds RC, Taylor PA, Glen DR (2023). Quality control practices in FMRI analysis: Philosophy, methods and examples using AFNI. Front. Neurosci. 16:1073800. doi: 10.3389/fnins.2022.1073800

So, these are things we learned we should check for, and we have a tool in (AP) that does so now. In your dataset's case, there weren't any found apparently.

To see an example of them in action, you can check out this recent MB post:


And here's another example of what variance line warnings can be pointing out:


thank you for your answer!!
And I would like to ask what is wrong with this result? What result counts as noise?


I suspect that that is a false positive for variance lines. Basically, there are just patches of high-variability there---we try to find only narrow lines of high variances, but larger regions can also get identified/highlighted, as well.

That is why we only list it as a warning and also show pictures, so people can check for sure. Just note that there is some inhomogeneity in the variance in your EPI. It doesn't look like the particular kind of variance line issue.

One additional thing to note: the high variance lines can also have the manifestation of affecting seedbased correlation outside the strict vertical line. This was observed in several datasets in the Reynolds et al. QC paper, cited above. In Fig 10, you can see some of what this looks like:

It will be easiest to run InstaCorr directly within the APQC HTML if you have a modern version of AFNI/, with the local server runnable through, as described in this OHBM-2023 poster:
"Modern" in this sense means updated since mid-August 2023.


Ok!Thank you for your helpful answers!