Hi, Brady-
Cool, thanks.
Re 1) That must be an input dataset you have included the 3dinfo data for? The mask+underlay images you show appear to have much, much higher resolution (and a spatially isotropic one, too). It is great to have the initial resolution information, but I suspect that the processing pipeline included a very large upsampling factor, then? Frankly, that output looks like an antomical resolution, or higher, consistent with.
The 2.75x2.75x5.00 mm**3 voxels are about 2.5 times the size of the ones I listed, so the similar whole brain mask would have less than 30k voxels, if made from that. Again, the images you show appear to have very, very high spatial resolution. I wonder if part of the reason for upsampling is try to counter some of the anisotropicity of the voxels?
While some software upsample highly by default, I am not quite sure what purpose it would serve, esp. to this high a resolution---upsampling won't add information, and it will greatly increase computational time and disk space storage. Is there a particular analysis that would require this? It is surprising to me.
Re 4) Hmm, if you had a one-to-one replacement files for each item in the AP results directory that goes into the APQC HTML, that might be possible to kinda retrofit with the generating script, but I doubt that can be done very easily... Do you have a lot of intermediate/output files from processing still? Frankly, setting up a new AP command would likely be faster, because you just need the input data. But I understand that might not be a popular choice.
That mask does show quite reasonable coverage, indeed. (And using any anatomical reference volume to define a final space is OK, in terms of processing setup and execution).
Re 5) Sounds good.
--pt
