Investigation of the effect of physiological factors on resting-state and task-based functional connectivity

Understanding the brain’s functional network through functional connectivity (FC) is crucial for gaining deeper insights into brain functional mechanism and identifying a potential biomarker for diagnosing neurological disorders. Despite the development of various FC measures, their reliability under different conditions remains under-explored. Moreover, physiological noise can obscure true neural activity, and accordingly, introduce errors into FC patterns. This issue necessitates further investigation. In this study, we evaluate and compare the performance of various methods using Local Field Potential and Blood-Oxygen-Level-Dependent signals across different conditions. We also examine the impact of physiological artifacts on BOLD-FC results. Our comprehensive assessment covers multiple modalities of brain signals, diverse task paradigms, and varying noise levels. Our findings reveal that while Granger Causality-based methods exhibit significant limitations, particularly with BOLD data, multivariate techniques (e.g. partial correlation) demonstrate greater robustness in distinguishing between different types of connections within the network. Notably, our results indicate that physiological artifacts substantially affect FC values, leading to erroneous connectivity estimates, especially with bivariate methods. This research offers a foundational analysis of the effects of physiological artifacts on FC results and provides valuable insights for future studies.