Quantitative comparison of resting-state functional connectivity derived from fNIRS and fMRI: a simultaneous...

The feasibility of functional near-infrared spectroscopy (fNIRS) to assess resting-state functional connectivity (RSFC) has already been demonstrated. However the validity of fNIRS-based RSFC has rarely been studied. In the present study, fNIRS and fMRI data were simultaneously acquired from 21 subjects during the resting state. After the spatial correspondence was established between the two imaging modalities by transforming the fMRI data into fNIRS measurements space, the index of Between-Modality-Similarity (BMS) of RSFC was evaluated across multiple spatial scales. First, the RSFC between the bilateral primary motor ROI was quite similar between fNIRS and fMRI for all the subjects (BMS(ROI) = 0.95 ± 0.04 for HbO and BMS(ROI) = 0.86 ± 0.13 for HbR). Second, group-level sensorimotor RSFC maps (0.79 for HbO and 0.74 for HbR) showed higher between-modality similarity than individual-level RSFC maps (0.48 ± 0.16 for HbO and 0.41 ± 0.15 for HbR). Finally, for the first time, we combined fNIRS and graph theory to investigate topological properties of resting-state brain networks. The clustering coefficient (C(p)) and characteristic path length (L(p)) which are the most important network topological parameters, both showed high between-modality similarities (BMS(Cp) = 0.90 ± 0.03 for HbO and 0.90 ± 0.06 for HbR; BMS(Lp) = 0.92 ± 0.04 for HbO and 0.91 ± 0.05 for HbR). In summary, the converged results across all the spatial scales demonstrated that fNIRS is capable of providing comparable RSFC measures to fMRI, and thus provide direct evidence for the validity of the optical brain connectivity and the optical brain network approaches to functional brain integration during resting state.