Objective: Human resting-state networks (RSNs) estimated from oxygenated (HbO) and deoxygenated hemoglobin (HbR) data exhibit strong similarities, while task-based studies show different dynamics in HbR and HbO responses. Such a discrepancy might be explained due to time-averaged estimations of RSNs. Our study investigated differences between HbO and HbR on time-resolved brain-wide coactivation patterns (CAPs).
Approach: Diffuse optical tomography was reconstructed from resting-state whole-head functional near-infrared spectroscopy data of HbR and HbO in individual healthy participants. Timeaveraged RSNs were obtained using the group-level independent component analysis. Time-resolved CAPs were estimated using a clustering approach on the time courses of all obtained RSNs. Characteristics of the RSNs and CAPs from HbR and HbO were compared.
Main results: Spatial patterns of HbR and HbO RSNs exhibited significant similarities. Meanwhile, HbR CAPs revealed much more organized spatial and dynamic characteristics than HbO CAPs. The entire set of HbR CAPs suggests a superstructure resulted from brain-wide neuronal dynamics, which is less evident in the set of HbO CAPs. These differences between HbO and HbR CAPs were consistently replicated in individual session data.
Conclusion: Our results suggest that human resting brain-wide neuronal activations are preserved better in time-resolved brainwide patterns, i.e., CAPs, from HbR than those from HbO, while such a difference is lost between time-averaged HbR and HbO RSNs.
Significance: Our results reveal, for the first time, HbR concentration fluctuations are more directly coupled with resting dynamics of brain-wide neuronal activations in human brains.