Analysis Of Driver's Cerebral Functional Network In Fatigue Based On Near-infrared Spectroscopy

With the rapid development of electronic control technology,data cloud technology and mobile communication technology,more and more driving assistant systems are designed for driving safety.While the design of auxiliary system mainly focus on easy handling and convenience,with little consideration of the additional mental consumption and whether the design can alleviate driving fatigue is not yet known.This study aims to establish a brain functional network model for driver to assess fatigue level by calculating various parameters.In this study,near-infrared spectroscopy was applied to detect the change of blood oxygen concentration in the prefrontal cortex(judgment),parietal lobe(motor area)and occipital lobe(visual area).Then,we establish the brain function model based on wavelet coherence and wavelet phase coherence and the graph-theory brain network model based on results of functional connectivity and effective connectivity.In the directed graph theory model,sampling areas were simplified as nodes and the weighted edge was defined by the coupling strength and coupling direction.Finally,calculate the fatigue evaluation parameters.Experiments were designed to verify the validation of the models:(1)Detect the concentration changes in prefrontal cortex and parietal lobe of 14 drivers using near-infrared instrument.After data pretreatment,wavelet coherence and phase coherence between channels were calculated,and then a brain functional connectivity model was established.In the first experiment,only the functional connectivity parameter was extracted and the statistical results revealed that,in intervals III-IV,the connectivity strength between brain regions decreased significantly with the prolongation of driving periods.Meanwhile,the driving performances parameters(Standard deviation of speed,Standard deviation of lateral position and Frequency of inappropriate crossing line)showed a significant decline of the vehicle control ability in fatigue state.The correlation results verified that brain functional connectivity model can assess driving fatigue effectively.(2)Detect the concentration changes in prefrontal cortex,parietal lobe and occipital lobe of 12 drivers using near-infrared instrument.Establish the graph-theory brain network model based on the values of wavelet phase coherence,coupling strength and coupling direction.Then parameters,clustering coefficient,shortest path,local efficiency and global efficiency were extracted to evaluate fatigue levels.Results showed that clustering coefficient reduced significantly,indicating that the degree of network modularity reduced.The significant extended shortest path and reduced global efficiency revealed that mental fatigue induced lower information transferring efficiency.Objective behavior parameters were also detected to verify the validity of the model,and the blink duration and blink frequency illustrated significant correlations with the fNMRS parameters.In conclusion,the brain functional connectivity model established in this paper can assess driving fatigue quantitatively,and the graph-theory brain network model can reveal the operational mechanism during driving and evaluate the fatigue level more reliably and fully.These models of fatigue assessment can be widely used in evaluating the rationality of the auxiliary driving system.