BA 47 mask alignment

I used the draw dataset plugin to make an anatomical BA 47 mask from the TT_N27+tlrc template. But–when I render it as overlay–it is off–more lateral and somewhat superior to actual TT_N27 anatomical. If I click “where am I” on the part that is lateral/not on top of anatomical–afni says I’m smack on BA 47. Any ideas–as I do not want to mis-apply a mask to my analyses. Thanks!


I’m not sure. I’ve just PM’ed you instructions on how to upload the dset, if you’d like me to take a look.


Hi again-

[EDIT: this suggestion does not actually give a data set, I was mistaken; this just outputs a text file listing which regions are being colorized (or are ‘None’). See the reply to this email for a better, automatic way to output the desired region mask directly from the commandline.]

Also note that if you are just literally re-drawing the BA-27 region, you can use AFNI’s “Atlas Colors” feature to just save a dset of the ROI. When you have the AFNI GUI open and a ulay dataset with “Talairach View”, you can right-click and select “–Apply Colors”. That will give you a dropdown menu (“TT Atlas Rendering”), and “Left Brodman area 27” and “Right Brodman area 27” are both there. You click on “Save” to save that ROI map to a dataset directly, as well as to view in the viewer. Doing so would save you having to draw the pre-existing ROI.



Probably the best way to output a region from a known atlas, rather than drawing something that exists, would be to dump it to a file directly using AFNI’s program “whereami”.

Here is an example of doing that for the Left BA 47:

whereami -mask_atlas_region TT_Daemon:left:Left_Brodmann_area_47 -prefix lh_ba47.nii

right BA 74 (using a slightly different syntax to above-- note how whereami figured out the hemisphere I wanted by just the 2nd part of the argument to -mask_atlas_region)

whereami -mask_atlas_region TT_Daemon:right:Brodmann_area_47 -prefix trighta.nii

or for both (NB: each ROI just has a value of 1 in every voxel, so they are not distinguishable numerically- if you want them separate, you could use 3dcalc to combine the above two masks, assigning each ROI a different integer label):

whereami -mask_atlas_region TT_Daemon::Brodmann_area_47 -prefix tboth.nii

To find out the name of the atlas and region you are looking for, you can dump all the atlas names+regions into a text file:

whereami -show_atlas_code > allatlas_rois.txt

and then search through. SEe the whereami help file for syntax of arguments (and to replace spaces with underscores).


The Talairach daemon atlas is not defined for the TT_N27 dataset specifically; it is only a rough approximation. The atlas was derived from histological slices from over fifty years ago, and there is no MRI nor any other kind of volume available that matches the regions specifically. Additionally, there are a number of discontinuities and errors in that atlas. The N27 dataset was put into Talairach space via a manual stereotaxic transformation. The Brodmann areas, in particular, make this problem more obvious because they are essentially only a voxel thick and follow the original contours of the individual half-brain from which they were derived.

Thanks Daniel for your response! Realizing the limitations of the Talairach atlas, and everyone’s time–is there a way to fix this/nudge them so they would be correct? At least the BA ones on edge of cortex that need a slight nudge to be on brain? As for AFNI folks in the universe, this not only would help ROI analyses, but it affects interpretation of the awesome WHEREAMI function. Thanks to you, Paul, Bob, and all @AFNI-central! -Dan