Atlas-free Brain Network Transformer

2025-09-30

AI Summary

Conventional brain network analyses relying on fixed atlases suffer from inter-subject spatial misalignment, regional functional heterogeneity, and atlas selection bias, limiting the reliability and interpretability of network representations. Method: We propose a novel atlas-free, individualized brain network modeling framework: (1) generating subject-specific functional parcellations directly from resting-state fMRI; (2) constructing voxel-level ROI-to-voxel functional connectivity features; and (3) designing a brain-network-adapted Transformer architecture to learn robust, comparable individual embeddings. Contribution/Results: By eliminating reliance on predefined atlases, our approach overcomes inherent limitations of atlas-based methods, significantly enhancing model interpretability, generalizability, and stability. Evaluated on sex classification and brain-age prediction, it outperforms state-of-the-art methods—including Elastic Net, BrainGNN, and Graphormer—demonstrating superior performance and establishing a new paradigm for personalized, precise neuroimaging analysis.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Humans and AI: Brain-Sensing and Analysis
• Computer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Current atlas-based approaches to brain network analysis rely heavily on standardized anatomical or connectivity-driven brain atlases. However, these fixed atlases often introduce significant limitations, such as spatial misalignment across individuals, functional heterogeneity within predefined regions, and atlas-selection biases, collectively undermining the reliability and interpretability of the derived brain networks. To address these challenges, we propose a novel atlas-free brain network transformer (atlas-free BNT) that leverages individualized brain parcellations derived directly from subject-specific resting-state fMRI data. Our approach computes ROI-to-voxel connectivity features in a standardized voxel-based feature space, which are subsequently processed using the BNT architecture to produce comparable subject-level embeddings. Experimental evaluations on sex classification and brain-connectome age prediction tasks demonstrate that our atlas-free BNT consistently outperforms state-of-the-art atlas-based methods, including elastic net, BrainGNN, Graphormer and the original BNT. Our atlas-free approach significantly improves the precision, robustness, and generalizability of brain network analyses. This advancement holds great potential to enhance neuroimaging biomarkers and clinical diagnostic tools for personalized precision medicine.

Problem

• Overcoming limitations of fixed brain atlases in network analysis
• Eliminating spatial misalignment and functional heterogeneity issues
• Improving reliability and interpretability of brain connectivity studies

Innovation

• Individualized brain parcellations from fMRI data
• ROI-to-voxel connectivity in standardized feature space
• Transformer architecture for comparable subject embeddings