The utilization of a fNIRS system to investigate brain function under different cognitive conditions: from a single modal to a multimodal system

Abstract

Since the first study on cerebral oxygenation in 1985, fNIRS system has been widely applied on various research areas, from the basic understanding of the brain to the complicated applications, and on various subjects, from infants to normal adults and to brain disorder elderly. Though fNIRS has been drawing attention from many people and thousands of papers related to this technique have been published, currently, it is still just being used in the research. The reason which makes fNIRS far away from being used in the clinic is that it has a low spatial resolution and it cannot measure deep brain signal. On the other hands, the low temporal resolution hampers its capability to be used in many real-life and online applications. However, fNIRS is simple, portable, and especially cost effective. Hence, further investigation on this technique may help to explore its appropriate applications in both the clinic and daily life. This work examines the potential of the fNIRS system as a single modal and also when combined with the EEG system to study the cognitive functions in different conditions and with different subject groups. Firstly, the single fNIRS system was used to identify the biomarker of the mild cognitive impairment (MCI) patients. Secondly, with the help of the EEG system, the fNIRS system was utilized to explore the cerebral functional connectivity in the rest and sleep states. Lastly, the fNIRS system was combined with the EEG system to find the physiological signals that can be used to predict driver drowsiness. In chapter 1, a brief description of fNIRS, the principle of this technique and how to calculate the cerebral hemodynamic responses from the measured light intensity, is introduced. In addition, current applications of this technique on the study of the cognitive functions are presented. In chapter 2, we examine the brain functional connectivity in both normal, healthy elders (HC) and mild cognitive impairment patients (MCI) to identify a typical feature to distinguish these two groups. A homemade four-channel fNIRS system is employed to measure the hemodynamic responses in the subjects’ prefrontal cortex during a resting state, an oddball task, a 1-back task, and a verbal fluency task. The brain connectivity is calculated as the Pearson correlation coefficients among fNIRS channels. The results reveal that MCI presents a significantly weaker connection in the inter-hemispheric connectivity than the left and right hemisphere connectivity. In addition, the connectivity in MCI exhibits a substantial decline along the performance time. Furthermore, comparison between MCI and HC connectivity establishes a statistically higher connection in the right hemisphere connectivity of MCI during the oddball task. These findings demonstrate potential of fNIRS as a promising avenue to study MCI brain functional connectivity. In chapter 3, we investigate the brain functional connectivity in the rest and sleep states. We collected EEG, EOG, and fNIRS signals simultaneously during rest and sleep phases using a commercial LABNIRS system. The rest phase was defined as a quiet wake-eyes open (w_o) state, while the sleep phase was separated into three states; quiet wake-eyes closed (w_c), non-rapid eye movement sleep stage 1 (N1), and non-rapid eye movement sleep stage 2 (N2) using the EEG and EOG signals. The fNIRS signals were used to calculate the cerebral hemodynamic responses (oxy-, deoxy-, and total hemoglobin). We grouped 133 fNIRS channels into five brain regions (frontal, motor, temporal, somatosensory, and visual areas). These five regions were then used to form fifteen brain networks. A network connectivity was computed by calculating the Pearson correlation coefficients of the hemodynamic responses between fNIRS channels belonging to the network. The fifteen networks were compared across the states using the connection ratio and connection strength calculated from the normalized correlation coefficients. Across all fifteen networks and three hemoglobin types, the connection ratio was high in the w_c and N1 states and low in the w_o and N2 states. In addition, the connection strength was similar between the w_c and N1 states and lower in the w_o and N2 states. Based on our experimental results, we believe that fNIRS has a high potential to be a main tool to study the brain connectivity in the rest and sleep states. In chapter 4, we introduce a new approach, a combination of EEG and NIRS, to detect driver drowsiness. EEG, EOG, ECG, and NIRS signals have been measured during a simulated driving task, in which subjects underwent both awake and drowsy states. Heart rate, blinking rate, alpha band power, and eye closure were used to identify each subject’s condition. EEG band power and hemodynamic responses were investigated in the awake and drowsy states. The oxy-hemoglobin change and the beta band power in the frontal lobe were found to differ the most significantly between the two states. In addition, the time course of these two parameters correspond well to the awake-drowsy transition. As a result, the beta band power and oxy-hemoglobin concentration change were used to derive a drowsiness detection index, which enabled the combined EEG/NIRS system to predict drowsiness earlier than a camera based method. From our study, we found that the fNIRS system itself can provide adequate information to study the cognitive functions of the brain. However, the additional information from the EEG system helps to add up the functions that are currently not available to the fNIRS system such as the sleep stage verification and also it helps to increase the accuracy such as the classification accuracy.