Network-Aware Bilinear Tokenization for Brain Functional Connectivity Representation Learning

2026-05-13

AI Summary

Existing approaches to parcellating brain functional connectivity often overlook the modular organization of large-scale brain networks, leading to representations that are misaligned with established neuroscientific priors. This work proposes NERVE, a novel framework that explicitly incorporates brain network priors into the parcellation process. By leveraging an anatomy-informed atlas, NERVE partitions the functional connectivity matrix into within-network and between-network blocks and introduces a structured bilinear factorization mechanism to embed heterogeneous connectivity patterns. This design preserves network identity while reducing parameter complexity from quadratic to linear. Integrated with a masked autoencoder, NERVE enables network-aware self-supervised learning. Evaluated on three major developmental cohorts—ABCD, PNC, and CCNP—NERVE substantially improves prediction of behavioral and psychopathological traits and yields representations that exhibit greater stability and cross-cohort generalizability.

Abstract

Masked autoencoders (MAEs) have recently shown promise for self-supervised representation learning of resting-state brain functional connectivity (FC). However, a fundamental question remains unresolved: how should FC matrices be tokenized to align with the intrinsic modular organization of large-scale brain networks? Existing approaches typically adopt region-centric or graph-based schemes that treat FC as structurally homogeneous elements and overlook the large-scale network brain organization. We introduce NERVE (Network-Aware Representations of Brain Functional Connectivity via Bilinear Tokenization), a self-supervised learning framework that redefines FC tokenization by partitioning FC matrices into patches of intra- and inter-network connectivity blocks. Unlike image-based MAE, where fixed-size patches share a common tokenizer, FC patches defined by network pairs are heterogeneous in size and correspond to distinct functional roles. To resolve this problem, NERVE embeds FC patches through a novel structured bilinear factorization. This formulation preserves network identity and reduces parameter complexity from quadratic to linear scaling in the number of networks. We evaluate NERVE across three large-scale developmental cohorts (ABCD, PNC, and CCNP) for behavior and psychopathology prediction. Compared to structurally agnostic MAE variants and graph-based self-supervised baselines, the proposed network-aware formulation yields more stable and transferable representations, particularly in cross-cohort evaluation. Ablation studies confirm that the proposed bilinear network embedding and anatomically grounded parcellation are critical for performance. These findings highlight the importance of incorporating domain-specific structural priors into self-supervised learning for functional connectomics.

Problem

• functional connectivity
• tokenization
• brain networks
• modular organization
• self-supervised learning

Innovation

• network-aware tokenization
• bilinear factorization
• functional connectivity
• self-supervised learning
• brain parcellation