Research map / Label-efficient and robust learning / Recognition and long-tail
A Hierarchical Tokenization Framework for Voxel-Level fMRI Representation Learning
Karan Singh, Yue Zhao, Mohammadhassan Abbasi, Ehsan Adeli
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes HBAR, a hierarchical tokenization framework that organizes voxel-level resting-state fMRI into sub-parcel voxel groups, parcels, and functional networks for compact multi-scale representation learning.
Problem
Voxel-level fMRI is high-dimensional, and conventional parcel-based representations discard fine-grained spatial information; a compact yet detailed representation is needed for neuroimaging learning.
Approach
A hierarchical tokenization over resting-state fMRI spanning sub-parcel voxel groups, parcels, and large-scale functional networks, giving about 415x compression while retaining fine-grained spatial detail beyond parcel-level methods.
Key results
- HBAR achieves approximately 415x compression of voxel-level resting-state fMRI while preserving fine-grained spatial information.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use atlas-based hierarchy instead of flat tokenizer at 415x | HBAR-415 · HCP-YA reconstruction (415x) | parcel correlation: 0.691 -> 0.746 (vs flat RVQ) | Improved |
| use hierarchical HBAR instead of MS-VQVAE at 415x for voxel reconstruction | HBAR-415 · HCP-YA reconstruction (415x) | voxel correlation: 0.427 -> 0.405 (MS-VQVAE vs HBAR) | Hurt |
| use hierarchy instead of flat BSQ at 52x | HBAR-52 · HCP-YA reconstruction (52x) | voxel and network correlation: voxel 0.529 vs 0.526; network 0.968 vs 0.965 | No clear effect |
| use hierarchy instead of flat BSQ at 52x | HBAR-52 · HCP-YA reconstruction (52x) | parcel correlation: 0.890 -> 0.920 | Improved |
| keep anatomical chunk-to-parcel grouping instead of shuffling (ablation) | HBAR-52 · HCP-YA reconstruction (52x) | parcel correlation: 0.886 -> 0.920 | Improved |
| keep anatomical grouping instead of shuffling (ablation) | HBAR-52 · HCP-YA reconstruction (52x) | voxel and network correlation: voxel 0.523 vs 0.526; network 0.966 vs 0.965 | No clear effect |
| use fixed atlas pooling instead of learned cross-attention pooling (ablation) | HBAR-52 · HCP-YA reconstruction (52x) | parcel correlation: 0.876 -> 0.920 | Improved |
| use fixed atlas pooling instead of soft alignment pooling (ablation) | HBAR-52 · HCP-YA reconstruction (52x) | parcel correlation: 0.796 -> 0.920 | Improved |
| HBAR instead of flat tokenizers for linear-probe demographics at 415x | HBAR-415 · HCP-YA age/sex linear probe (415x) | AUROC: sex 0.701 vs 0.722 (RVQ); age 0.482 vs 0.607 (BSQ) | No clear effect |
| HBAR instead of flat tokenizers for AR readout at 415x | HBAR-415 · HCP-YA AR readout (415x) | AUROC: age 0.506 -> 0.584; sex 0.556 -> 0.638 (vs flat BSQ) | Improved |
| HBAR instead of flat BSQ for sex AR readout at 52x | HBAR-52 · HCP-YA sex AR readout (52x) | AUROC: 0.640 -> 0.792 | Improved |
| add temporal-difference loss terms (ablation) | HBAR-52 · HCP-YA 52x dev ablation | voxel recon MSE: 0.810 -> 0.673 | Improved |
Closest papers by meaning
- Unsupervised Representation Learning of Brain Activity via Bridging Voxel Activity and Functional Connectivity (ICML 2024 · similarity 0.87)
- Brain-Semantoks: Learning Semantic Tokens of Brain Dynamics with a Self-Distilled Foundation Model (ICLR 2026 · similarity 0.87)
- Brain Harmony: A Multimodal Foundation Model Unifying Morphology and Function into 1D Tokens (NeurIPS 2025 · similarity 0.86)
- FlatClip: Reusing Image Foundation Models for fMRI Representation Learning via Cortical Flatmaps (NeurIPS 2026 · similarity 0.86)
- DCA: Graph-Guided Deep Embedding Clustering for Brain Atlases (NeurIPS 2025 · similarity 0.85)
- BrainLM: A foundation model for brain activity recordings (ICLR 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
