Abstract
Near-Infrared Spectroscopy (NIRS) measures cerebral blood flow activated by neural activities as changes of Oxy-hemoglobin concentrations by using near-infrared light. NIRS signals contain not only neural activities, but also cardiovascular fluctuations, such as respiratory sinus arrhythmia (RSA) and Mayer wave related sinus arrhythmia (MWSA). Usually, these noises are cancelled by measuring same tasks repeatedly and averaging them. However, this averaging method needs many tasks and long-time measurements that lead to subjects' stress. In this study, we proposed a new method to remove cardiovascular fluctuations from NIRS signals by measuring cardiovascular signals simultaneously and using general linear model (GLM) method. First, we estimated powers of RSA and MWSA in NIRS signals by a respiratory-control experiment. Then, we evaluated the accuracy of the proposed method using computationally generated data. Finally, we demonstrated the effectiveness of the proposed method in a working memory experiment, which is one of psychological experiment. This result shows our method enables measuring NIRS signals not only more shortly, but also more accurately.