Add mouse brain atlas

[araikes]

Hi @vmallela0,

How hard would it be to add a mouse brain atlas? Specifically, I'm thinking about the Dorr and DSURQE atlases (https://wiki.mouseimaging.ca/display/MICePub/Mouse+Brain+Atlases)

Let me know what you think.

[araikes]

Bump for visibility @vmallela0 or @razvanmarinescu

[vmallela0]

sorry for the late reply @araikes. if the mouse atlas comes in .ply (or a similar format) files, it wouldn't be difficult at all but if we need to run segmentation on a nifti mouse atlas, it's going to be extremely difficult since softwares like free surfer don't have the capabilities to segment mice brains.

[araikes]

So a massive challenge :D

Understood. Thanks.

[razvanmarinescu]

@araikes, actually I think it's possible, cause you can just manually segment one scan only. Those segmentations can then be used to create those ply objects for blender.

If you, or a labmate/collaborator, can create that segmentation, that is enough. And it has to follow those atlases. Also, can you check if they provide a segmentation template? Most studies generally do.

[araikes]

@razvanmarinescu,

I have a NIFTI template image and a multi-label atlas image. Are you suggesting that this is feasible if I simply break apart the atlas into individual segmentations and use the template?

[razvanmarinescu]

@araikes yes, exactly. you only need the segmentation files (don't need the original NIFTI). You just need to break the multi-label segmentation apart into individual segmentations, then convert each of them to the .ply format, which is a polygon mesh format. @vmallela0 how did you convert the voxelized segmentations to the ply format?

[araikes]

@razvanmarinescu,

Does it need to be in Allen Brain space (the one for the existing mouse brain painter) or does it just create the brain out of the parcellations colored to whatever the background color is and then add the color for those that aren't background?

[razvanmarinescu]

it doesn't need to be in Allen Brain space. But the 3D models for each ROI cannot be centered all in the origin, they need to be placed in the right location, so as to create a coherent brain.

[razvanmarinescu]

But with those atlases you mentioned, you might get them already translated & rotated, so you might not need to do anything special on that

[araikes]

It'll definitely create a coherent brain just using the labeled ROIs.

@vmallela0, pinging you once more for input... Only because I'm trying to get this done this week for a progress report that would benefit from the really good images from brainpainter.

[vmallela0]

@araikes when I transformed the ply files from the allen atlas I manually fit them together in order to make the full brain through the blender software. when you segment your atlas, though, you might need to scale it to fit into the BrainPainter renderer. I can assist with the scaling if you can send me the ply files

[araikes]

@vmallela0, so contrary to @razvanmarinescu's comment, you didn't convert voxelized segmentations to PLYs?

[vmallela0]

@araikes when we did the mice brain atlas, we retrieved the already converted voxelized segmentations from the allen institute and ran transformations on them in order to be compatible with the brainpainter software

[araikes]

Ok @vmallela0,

I think I have this done and it ended up being much easier than I thought (though I'm not sure if the level of smoothness in the meshes will work). I used ITK Snap to export the segmentation to individual meshes in STL format. I then used Blender to convert the STL files to PLYs.

I opened all of the PLY files together in Blender and they appear to converge to form a mouse brain. I have two different atlases that I'd like to be able to use (different levels of granularity) and each has the associated PLY files. How can I send them to you?

[vmallela0]

perfect! you can send them to me at vmallela@csail.mit.edu. I'll get those up to brainpainter asap and i'll ping you here once they're uploaded on the software. Thanks!