Regional Brain Entropy, Brain Network and Structural-Functional Coupling in Human Brain

Abstract

Brain entropy (BEN) indicates the irregularity, unpredictability and complexity of brain activity. In healthy brains, resting-state fMRI-based regional BEN (rBEN) distribution has been shown to have potential relationships with functional brain networks. However, the relationship between rBEN and structural and functional networks, as well as how rBEN facilitates the coupling between structural and functional networks remains unclear. Additionally, network dimensionality reduction methods, such as the use of connectome gradients, have become popular in recent years for explaining the hierarchical architecture of brain function, and the sensorimotor-association (S-A) axis, as a major axis of hierarchical cortical organization, exerts a widespread influence on both brain structure and function. However, how this functional hierarchy affects the relationship between rBEN and brain networks remains unclear. In this study, we systematically examine the relationship between rBEN and both structural and functional networks, including BOLD-based functional networks and MEG-based functional networks. We also assess the impact of the BEN and network gradients, as well as the influence of the S-A axis, on the relationship between rBEN and brain networks. Our results reveal a negative correlation between BEN and the network efficiency of both structural and BOLD-based functional networks, as assessed by average connection strength, degree centrality, and local efficiency. Additionally, BEN shows a negative correlation with the coupling between structural and BOLD functional networks. In contrast, the relationship between BEN and MEG functional network efficiency varies from negative to positive across different frequency bands, with a shift from negative to positive correlation observed in the coupling between BEN and MEG-functional networks. Moreover, the coupling between BOLD and MEG functional networks is positively correlated with BEN. The relationship between BEN and network gradients is complex, and no consistent patterns were observed. Importantly, the relationship between BEN and brain networks is influenced by the S-A axis, and the connection between BEN and networks is further modulated by changes in cytoarchitectural organization. These results are consistent with the commonly observed relationship between elevated rBEN and impaired networks in psychiatric disorders. The negative correlation between BEN and the efficiency of both structural and BOLD functional networks, as well as their coupling, may suggest that lower rBEN is associated with higher information processing potential. Additionally, the positive correlation between BEN and high-frequency MEG functional networks, as well as the coupling between BOLD and MEG functional networks, may indicate that brain processes related to information acquisition and integration could increase rBEN. In summary, this complex relationship may reflect a dynamic process in which the brain continuously acquires external information for integration (increasing entropy) and internalizes it (decreasing entropy) as part of its information-processing mechanism in different timescales.