Tutorials#
End-to-end, narrative walkthroughs on real data — the place to learn by doing. Each notebook is rendered with its committed output (figures and all), so you can read the whole story without running anything; clone the repo to execute them yourself.
How these are built
The notebooks live in the repository’s top-level tutorials/ folder. The doc
build copies them into this section and renders them without re-executing
(nb_execution_mode = "off"), because several need real surfaces or a GPU. See
the Contributing page to rebuild them from their generators.
- 01 · The Laplace–Beltrami operator: the engine under SpectralBrain
- 02 · Reading real brains: the I/O layer
- 03 · ShapeDNA: hearing the shape of a hippocampus
- 04 · The Heat Kernel Signature: diffusion as a local probe
- 05 · Wave Kernel Signature & Global Point Signature
- 06 · Point clouds & white-matter tracts
- 07 · Functional maps & shape distances
- 08 · Cohorts & vertex-wise statistics
- 09 · Effect sizes, classification, harmonization
- 10 · Bayesian spectral analysis & visualization (capstone)
The arc#
Laplace–Beltrami operator — the eigenproblem, hands-on.
Reading real brains (I/O) — FreeSurfer, GIfTI, NIfTI, HippUnfold, TractSeg.
ShapeDNA — the global fingerprint and shape distance.
Heat Kernel Signature — multiscale per-vertex geometry.
Wave Kernel & GPS — band-pass signatures and spectral embedding.
Point clouds & tracts — volumetric segmentations and white-matter bundles.
Functional maps & distances — cross-shape correspondence and metrics.
Cohorts & vertex-wise stats — group loading, FWE permutation, FDR, TFCE.
Effect sizes, classification & harmonization — ComBat/ComBat-GAM, AUC.
Bayesian & visualization — PyMC models and publication figures.