Reproduce “A Hierarchical Tokenization Framework for Voxel-Level fMRI Representation Learning” 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

Karanpartap Singh, Yue Zhao, Mohammad H. Abbasi, Ehsan Adeli

HBAR compresses resting-state fMRI voxel time series into discrete codes organized at voxel-group, parcel, and network scales. Its atlas-based hierarchy preserves coarse-scale signal summaries while retaining voxel-level inputs. In HCP-YA, at approximately 415× code-payload compression, parcel and network correlations were 0.746 ± 0.003 and 0.895 ± 0.002, exceeding rate-matched flat RVQ; this does not establish full time-course preservation.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.