Optimization of wavelet coherence analysis as a measure of neural synchrony during hyperscanning using functional near-infrared spectroscopy

Significance: The expanding field of human social interaction is enabled by functional near-infrared spectroscopy (fNIRS) that acquires hemodynamic signals during live two-person interactions. These advances call for development of methods to quantify interactive processes. Aim: Wavelet coherence analysis has been applied to cross-brain neural coupling. However, fNIRS-specific computations have not been explored. This investigation determines the effects of global mean removal, wavelet equation, and choice of oxyhemoglobin versus deoxyhemoglobin signals. Approach: We compare signals with a known coherence with acquired signals to determine optimal computational approaches. The known coherence was calculated using three visual stimulation sequences of a contrast-reversing checkerboard convolved with the canonical hemodynamic response function. This standard was compared with acquired human fNIRS responses within visual cortex using the same sequences. Results: Observed coherence was consistent with known coherence with highest correlations within the wavelength range between 10 and 20 s. Removal of the global mean improved the correlation irrespective of the specific equation for wavelet coherence, and the oxyhemoglobin signal was associated with a marginal correlation advantage. Conclusions: These findings provide both methodological and computational guidance that enhances the validity and interpretability of wavelet coherence analysis for fNIRS signals acquired during live social interactions. (C) The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License.