Thread by @raamana_

#neuroimaging folks, we really need a high-quality #python package or Class to read and work with volumetric and surface atlases. Atlases are central to our work and are still not first class objects!!

I proposed aspects of this few years ago for nilearn: github.com/nilearn/nilear…

I’m trying to deal with different atlases for #graynet : github.com/raamana

And it’s not as clean and beautiful as I’d like it to be so far.

Happy to collaborate and work with others to achieve niatlas package.

Atlases shouldn’t be handled on a per-tool or per-ecosystem basis, and we really need a nibabel-like tool dedicated to atlases, with high interoperability across various popular tools.

There is already bits and pieces of this work in different places e.g. lots of atlases in MNI space from @leaddbs: lead-dbs.org/helpsupport/kn…

scripts to transfer atlases in fsaverage space to subject space: github.com/faskowit/multi…

FS parcellation fragmenter
github.com/miykael/parcel…

Lot of Brain signature patterns, atlases of regions, and meta-analysis masks from @torwager's lab (although in Matlab): github.com/canlab/Neuroim…

And many others that I am not aware of.

I must mention pysurfer which is great for visualizations for atlases, esp related Freesurfer: pysurfer.github.io/auto_examples/…