BrainCSD: A Hierarchical Consistency-Driven MoE Foundation Model for Unified Connectome Synthesis and Multitask Brain Trait Prediction

2025-11-07

AI Summary

Clinical applications of functional connectivity (FC) and structural connectivity (SC) are hindered by high acquisition costs, complex preprocessing, and frequent modality incompleteness. To address these challenges, we propose a hierarchical consistency-driven Mixture-of-Experts (MoE) foundation model. Our method integrates three specialized MoE modules—region-specific, dynamic activation, and network-aware—combined with contrastive consistency learning and structural prior regularization to explicitly model cross-modal (FC/SC) and multi-scale (regional/network/dynamic) consistency. Under modality-incomplete conditions, the model achieves 95.6% accuracy in mild cognitive impairment (MCI) classification, with FC and SC synthesis RMSEs of 0.038 and 0.006, respectively; brain age prediction MAE of 4.04 years; and MMSE score estimation MAE of 1.72 points. These results demonstrate substantial improvements in joint multi-task modeling capability and generalization robustness.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Multi-modal Vision
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Functional and structural connectivity (FC/SC) are key multimodal biomarkers for brain analysis, yet their clinical utility is hindered by costly acquisition, complex preprocessing, and frequent missing modalities. Existing foundation models either process single modalities or lack explicit mechanisms for cross-modal and cross-scale consistency. We propose BrainCSD, a hierarchical mixture-of-experts (MoE) foundation model that jointly synthesizes FC/SC biomarkers and supports downstream decoding tasks (diagnosis and prediction). BrainCSD features three neuroanatomically grounded components: (1) a ROI-specific MoE that aligns regional activations from canonical networks (e.g., DMN, FPN) with a global atlas via contrastive consistency; (2) a Encoding-Activation MOE that models dynamic cross-time/gradient dependencies in fMRI/dMRI; and (3) a network-aware refinement MoE that enforces structural priors and symmetry at individual and population levels. Evaluated on the datasets under complete and missing-modality settings, BrainCSD achieves SOTA results: 95.6% accuracy for MCI vs. CN classification without FC, low synthesis error (FC RMSE: 0.038; SC RMSE: 0.006), brain age prediction (MAE: 4.04 years), and MMSE score estimation (MAE: 1.72 points). Code is available in href{https://github.com/SXR3015/BrainCSD}{BrainCSD}

Problem

• Synthesizes functional and structural brain connectivity biomarkers
• Enables cross-modal consistency for missing neuroimaging data
• Supports multitask brain trait prediction and disease diagnosis

Innovation

• Hierarchical MoE model synthesizes brain connectivity biomarkers
• ROI-specific MoE aligns regional activations via contrastive consistency
• Network-aware refinement enforces structural priors and symmetry