Cross-correlation between HbO and HbR as an effective feature of motion artifact in fNIRS signal

The general linear model (GLM) as a standard model for fMRI analysis has been applied to fNIRS imaging analysis as well. The GLM is very likely to make false predictions for motion artifact in fNIRS signals. The temporal characteristics of normal cerebral hemodynamics are basically the opposite tendency of oxy-hemoglobin (HbO) and deoxy-hemoglobin (HbR). When motion artifact occurs, HbO and HbR are completely different from normal cases as the baseline changes. This paper presents a cross-correlation between HbO and HbR as a feature that can determine the dynamic noise of fNIRS. Since cross-correlation is a deterministic tool that is easy to calculate, it will be very useful for noise elimination if it is noted as a criterion of dynamic noise in fNIRS signals.