Concurrent multimodal imaging demonstrates that EEG-based excitation/inhibition balance reflects glutamate and GABA concentrations

• Published in: bioRxiv
• Main Authors: Cochrane, Aaron, Rosedahl, Luke, Tamaki, Masako, Watanabe, Takeo, Sasaki, Yuka
• Format: Journal Article Paper
• Language: English
• Published: United States Cold Spring Harbor Laboratory 10.06.2025
• Edition: 1.2
• Subjects: Neuroscience
• ISSN: 2692-8205; 2692-8205
• DOI: 10.1101/2025.03.31.646339

Neural excitation/inhibition (E/I) ratio is dynamically regulated on multiple timescales. Adaptive changes in E/I ratio can support healthy development, learning, and cognition, while disordered E/I ratio has been implicated in neurodevelopmental disorders, neurodegenerative disorders, and states of impaired vigilance. There has been growing interest in inferring E/I ratio from efficient and noninvasive measurements such as electroencephalography (EEG), and several algorithms have been proposed to estimate E/I ratio from EEG. Despite promising results, there has been a lack of validation studies testing the underlying neurochemical changes leading to increased or decreased EEG-based E/I ratio. Here, using concurrent EEG and magnetic resonance spectroscopy (MRS) for over an hour, we assessed which algorithm of EEG-based E/I ratio best matched with MRS-based E/I ratio in humans of both sexes. The MRS-based E/I ratio was obtained by the ratio of glutamate concentration to GABA concentration. We applied 10 candidate indices of EEG-based E/I ratio using four approaches in several spontaneous frequency bands. Uniquely, we quantified the associations between the EEG-based E/I ratio and MRS-based E/I ratio separately for between-subjects and within-subjects variations. We found that each EEG-based E/I algorithm showed reliable and positive associations with MRS-based E/I, and especially EEG-based E/I ratio in alpha band with a criticality theory based approach showed the best association to the MRS-based E/I ratio. While these associations were evident for between-subjects comparisons, they were quite weak for within-subjects comparisons. These results suggest that EEG-based E/I algorithms are likely to reflect, at least in part, relative concentrations of glutamate and GABA.