CortexBridge: Cortical Alignment of EEG Montages for Foundation Models

2026-10-01

AI Summary

This study addresses the cross-device transfer challenge in EEG foundation models caused by fixed channel vocabularies and restricted montage layouts. We propose a lightweight adapter method that formulates cortical alignment as a learnable, anatomically grounded routing mechanism. By fusing electrode coordinate embeddings with atlas information, this approach maps arbitrary montage configurations into a shared cortical latent space, enabling spatial alignment of heterogeneous data while keeping the foundation model frozen. Experimental results on the MOABB benchmark demonstrate substantial performance improvements, achieving average gains of up to 13.02%. These findings indicate that the proposed method effectively overcomes the limitations of conventional static channel constraints, facilitating robust cross-device generalization for EEG foundation models.

Abstract

Electroencephalography (EEG) foundation models are often pretrained with a fixed channel vocabulary or a limited set of montages, making transfer difficult when electrode layouts change. We propose CortexBridge, a lightweight adapter that combines EEG features with electrode and atlas coordinates to map arbitrary montages into a shared cortical latent space. Evaluated with three frozen foundation models on five brain-computer interface (BCI) datasets from the Mother of All BCI Benchmarks (MOABB), CortexBridge improves performance in 13 of 15 evaluations. The gains in balanced accuracy average 0.80% for EEGPT, 0.70% for LaBraM, and 3.26% for CBraMod, with a maximum gain of 13.02% on 12-class steady-state visual evoked potential (SSVEP) classification. Visualizations of the learned atlas representations reveal task-dependent spatial patterns, with SSVEP showing a more concentrated representation in the Yeo Visual network than auditory P300. These results establish cortical alignment as a learnable and anatomically grounded routing mechanism from heterogeneous EEG montages to pretrained foundation models.

Problem

• EEG foundation models
• montage transfer
• electrode layout
• cortical alignment
• brain-computer interface

Innovation

• Cortical Alignment
• EEG Foundation Models
• Lightweight Adapter
• Brain-Computer Interface
• Shared Latent Space