Interpreting results from SSWarper

Good morning,

Our lab is using SSWarper for the first time. We have been able to get it to work successfully after some glitches, and had a few questions about the output that we could not find from the --help or SSWarper page on the afni website. I am attaching the AM and MA images that were produced. I purposely chose a participant who seemed to have some issues with motion to see what the output would look like.

Could you clarify what the colors mean on these two images? There doesn’t seem to be a key anywhere to which color means what. Am I correct that it takes the MNI152 template (or whatever template you choose), and overlays the results onto that, which then accounts for the lines?

Also, how do you decide if the output from SSWarper is sufficient to move onto the next steps of pre-processing? And if the registration doesn’t seem sufficient, what steps are you supposed to take? We are trying to figure out how to assess quality of the output, how much “error” to accept in the colored lines, and what do do if the quality seems poor.

Thank you!

Hi, Sarah-

The lines are of the volume, made using 3dedge3-- they show where the contrast is changing a lot. The colorbar is Plasma, with yellow showing the largest changes in contrast (hence it is usually at the edge of the brain) and violet/purple showing the smaller differences.

The images are of the template (that you choose) and anatomical brain, yes, in the following representations:

AMsub007.jpg            = 3x3 snapshot image of the anatQQ.sub007.nii
                            dataset with the edges from the base template
                            overlaid -- to check the alignment;
  MAsub007.jpg            = similar to the above, with the roles of the
                            template and the anatomical datasets reversed.

Knowing how much variability is “OK” is tricky. It might even vary by study. What you probably don’t want, is having just one or two subjects with very different alignment patterns. How much is “very” (and how would that even be quantified?) is not known, unfortunately.

To look at these, you might consider opening all the images of one type (either AM* or MA*) and flipping through them like a movie or flipbook. Your eye will very quickly ascertain the amount of variability-- and likely any oddball alignments.

How to go back and fix an alignment depends on what is happening, I guess. If you have a brain you are particularly worried about, I suppose you could send the data and we could take a look for any particularly odd feature that might be causing issues. If you have non-human and/or non-adult data, them maybe using a different template base would be a place to start (e…g, the Haskins pediatric one, for children’s brains).


Thank you very much for your response! It is quite helpful.



Thanks again for your help with this. I had one more question–any recommendations on how to run multiple participants at once with the SSWarper script? We have a SLURM scheduler on our institute’s HPC that needs to have a few lines of command above to launch the required version netpbm as well.

Thank you!

Hi Paul,

I would like to ask a follow-up question of this thread. Do you have any suggestion on how to check the quality of warping lesioned brains using @SSWarper?