Reproduce “Fast Whole-Brain, Geometry-Aware Functional Alignment for Cross-Subject Decoding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Pierre-Louis Barbarant, Florent Meyniel, Bertrand Thirion
Researchers introduced SpectralOT, a functional alignment method leveraging Laplace-Beltrami eigenmodes to integrate cortical geometry with optimal transport for fMRI data. This approach achieves cross-subject decoding performance comparable to state-of-the-art methods while being over 30 times faster and simpler to parameterize.
Key insights
- Laplace-Beltrami eigenmodes effectively encode global cortical geometry, providing robust anatomical regularization that, when combined with functional data, enhances alignment.
- Linearly combining functional and geometric costs within a single Optimal Transport framework simplifies optimization and significantly reduces computational burden compared to complex joint minimization approaches.
- A balanced contribution of both anatomical and functional constraints (intermediate α values) yields superior inter-subject functional alignment compared to purely functional or anatomical approaches.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.