Spatial basis functions should be graded orthogonal maps, as opposed to discrete parcellations.
A handful are provided here for ease of use with CANlab tools, but for a more comprehensive or rigorous analysis you should look into NeuroMaps: https://netneurolab.github.io/neuromaps/usage.html.
Each sub-folder contains one set of orthogonal maps (a "basis") with
its NIfTI(s), contents_description.md, visualize_contents.m, and a
short methods write-up.
| Folder | Description | Loader keyword |
|---|---|---|
| margulies | Pilot — Margulies et al. 2016 first principal gradient of cortical organisation (unimodal → transmodal). CANlab volumetric build via registration fusion. | marg, transmodal, principalgradient, margfsl |
| transcriptomic_gradients | First three principal components of the Allen Human Brain Atlas gene-expression data (Burt et al. / Anderson et al.). | transcriptomic_gradients |
| hcp_91k | HCP 91k-grayordinate spatial bases (CIFTI). | — |
| hcp_groupICAs | HCP group-ICA spatial maps at multiple model orders. | — |
| mitochondrial_profile_maps | Mitochondrial-pathway profile maps (Mosharov / Picard 2025 Nature). | — |
See the docs README. Each folder retains any
existing README.md verbatim; the contents_description.md adds the
standardised overview / inventory / citation section and links back to
that README.