Perens stereotaxic mri

[PolarBean]

Here I added the Perens MRI stereotaxic atlas

Description

What is this PR

• Bug fix
• Addition of a new feature
• Other

References

#203

How has this PR been tested?

I ran the atlas generation script, and inspected the output.
I attempted to run the mesh checking however I am missing the correct drivers for libGL.

[alessandrofelder]

Thanks a lot for this @PolarBean
I've left some high-level comments for now - let me know what you think.

[alessandrofelder]

Looking good - thanks @PolarBean
Some more very minor comments.

[PolarBean]

@alessandrofelder reformatted to the new template and addressed your comments. Feel free to re-review when you get a chance.

[adamltyson]

@PolarBean is it possible to add a few more print statements, and ideally a download progress bar? It's hard to know what's going on when running this script.

[adamltyson]

Thanks for working on this @PolarBean.

I'm not exactly sure what files are packaged with this atlas, but presumably we should be packaging up multiple reference (template) images? If I understand correctly, the main reason to add this atlas would be to make available all the different reference images as registration targets?

It may also be worth packaging up multiple atlases, one for each coordinate space.

[PolarBean]

Thanks for working on this @PolarBean.

I'm not exactly sure what files are packaged with this atlas, but presumably we should be packaging up multiple reference (template) images? If I understand correctly, the main reason to add this atlas would be to make available all the different reference images as registration targets?

It may also be worth packaging up multiple atlases, one for each coordinate space.

This atlas only has one template image (MRI). It is in a different space to the perens LSFM atlas. It's valuable because it is in flat skull position and is therefore useful for stereotaxic surgery. It is also useful as a MRI registration target. I will make a seperate pull request for the most recent perens lsfm volume once they have updated those files on their server.

[adamltyson]

Ah, nvm. I assumed as it was a "multimodal" atlas framework, there were templates for LSFM & CT in the same space.

[PolarBean]

Yeah they have multiple templates in multiple spaces and they are linked by deformation matrices.
I've also developed https://github.com/Neural-Systems-at-UIO/CCF_translator to make transferring data between the spaces more user friendly.

[PolarBean]

If I understand correctly, the main reason to add this atlas would be to make available all the different reference images as registration targets?

Also one thing I'm curious about is what should our approach be to adding new resources to brainglobe. Should a new atlas provide some benefit in order to be added, or instead is just the existence of a resource enough to add it (even if it has been superseded by newer better resources)? I would tend towards wanting to add in everything, as it could be useful for say replicating past results or comparing atlas definitions of particular regions. But others may disagree.

[adamltyson]

I agree with you. As an example this atlas was added to allow for some pre-BrainGlobe analyses to be replicated with BrainGlobe tools.

If an atlas has obviously been replaced by another, then I would suggest we:

• Make sure this is noted on the website to ensure users are using the best atlas for them
• We don't prioritise adding it, and try to focus on the atlases with the biggest impact (which I assume will be the "better" ones)

[PolarBean]

@PolarBean is it possible to add a few more print statements, and ideally a download progress bar? It's hard to know what's going on when running this script.

No worries, I've added a download progress bar and a few more print statements. Let me know if it's what you had in mind

[adamltyson]

No worries, I've added a download progress bar and a few more print statements. Let me know if it's what you had in mind

Yes, thanks! A bit more printing (e..g just to say what's being downloaded might be useful too).

I haven't checked the packaged atlas in detail (I'll leave that to @alessandrofelder when he's back), but I noticed a couple of issues:

• The packaged atlas is in ars orientation (rather than asr). This would suggest that the orientation set in the script is wrong. This may be because load_any() is loading the raw image data, and not transforming the data according to the nifti metadata
• There are many more stuctures (840) than meshes (754). It's not always possible to create a mesh for every region, but nearly 100 missing is surprising. Not sure why this has happened though.

[PolarBean]

• The packaged atlas is in ars orientation (rather than asr). This would suggest that the orientation set in the script is wrong. This may be because load_any() is loading the raw image data, and not transforming the data according to the nifti metadata

I see, I'll have to recheck it. I thought it was unchanged from the previous perens lsfm in the apiwhich is listed as IAL and so I copied it from there. I will recheck it for this one and for the other pull requests I have.

[PolarBean]

• There are many more stuctures (840) than meshes (754). It's not always possible to create a mesh for every region, but nearly 100 missing is surprising. Not sure why this has happened though.

Hmm, I contacted Gubra about some problems I was facing when ingesting the LSFM volume (for which I will soon open a pull request). They fixed that problem and it seems they also updated the MRI annotation while they were at it. When I run this script I now get 838 meshes which seems much more respectable!

[PolarBean]

I read the documentation but Im a bit confused about differentiating between ars and als for symmetrical atlases. I put als as i thought left to right seemed most intuitive but I am a bit lost.

[adamltyson]

I read the documentation but Im a bit confused about differentiating between ars and als for symmetrical atlases. I put als as i thought left to right seemed most intuitive but I am a bit lost.

In principle that shouldn't make a difference.

[PolarBean]

A bit more printing (e..g just to say what's being downloaded might be useful too).

it actually only downloads a single file as all the atlas data is distributed in a single archive. I'll add print statements elsewhere though

[PolarBean]

ok actually i added one more print statement but it seems now every function has at least one print statement. I think I've addressed everything now but feel free to add more recommendations. I'll also update the template script with print statements so it should be included by default in the future.

[PolarBean]

I ran the validation script Alessandro sent me and it returned the following error

Failed validations:
('catch_missing_mesh_files', "Structures with IDs [545, 1051] are in the atlas, but don't have a corresponding mesh file.")
WARNING: Could not load the Qt platform plugin "xcb" in "" even though it was found.
WARNING: This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.

Available platform plugins are: eglfs, minimal, minimalegl, offscreen, vnc, webgl, xcb.

Aborted (core dumped)

I'll try to fix the xcb error but the missing structures are likely just due to those regions being too small in this atlas.

[PolarBean]

ok after resolving the difficulties the validation script ran and showed me this volume. It all looks as expected!

[alessandrofelder]

Thanks @PolarBean - I'll review next week!