麻醉中神经血管耦合估计与分析

Recent studies show that the analysis combining the electroencephalogram (EEG) and near infrared reflectance spectroscopy (NIRS) could reveal the neurovascular coupling of the brain. This provides a new perspective for studying the mechanisms of consciousness and the effect of anesthetic. The key issues in interpreting the mechanism of anesthesia is to develop a reliable neurovascular coupling function by studying the correlation of EEG and NIRS signals in both temporal and spatial domains. In this proposal, the concurrent anesthetic EEG/NIRS data are acquired from the rat experiments and clinical surgeries. Then, the following methods are used to analyze the data: The permutation mutual information and phase-amplitude coupling methods are employed to calculate the coupling strength of EEG and hemodynamic response; The generalized linear model is adopted to evaluate the causality of local hemodynamic and EEG in the spatial domain; The evolution map approach and spatial independent component analysis (ICA) are used to map the brain network and functional connectivity for EEG and NIRS data respectively. Next, a neurovascular coupling model modified by a first-order phase transition with the anesthetic effect is proposed to describe the transient dynamics of the induced unconsciousness, whose parameters are optimized by Bayesian inversion algorithm based on the experiments. And the model is used to generate surrogate EEG and hemodynamic signals to evaluate the algorithm's performance. Finally, in order to obtain an effective index of drug effect against different drug concentration in the multi-parameter analysis, an artificial neural network based on the group search optimization is used. Furthermore, the performance of the index is estimated by prediction probability. The work of this project is expected to provide new tools for studying the neurovascular coupling during anesthesia, as well contribute to the understanding of the mechanism of anesthesia. The outcomes will help to build the theoretical foundation of anesthesia and consciousness in a way.

最新研究表明，集成脑电和近红外成像技术获取的神经血管耦合信息为研究麻醉药物效应及意识机理提供了新的思路。研究可靠的脑电和近红外信息耦合估计和分析方法，发展有效的神经血管耦合指数，是药物效应和意识机理评估的关键问题。本项目首先基于大鼠实验和临床获取的脑电及近红外数据，采用排序互信息和相位-幅度耦合方法计算神经血管耦合强度，并基于广义线性模型描述两种信息的局部因果关系；引入进化图和空间独立分量分析方法评估脑区间两者的耦合强度和方向；其次结合一阶相变理论和药物代谢建立麻醉中的神经血管耦合模型，采用贝叶斯反演优化模型参数，使其产生替代脑电与血流动力学数据用于算法性能分析；最后针对不同药物浓度效应，以群搜索的人工神经网络优化效应指数，并采用预测概率评价指数性能。课题的开展将为研究麻醉下的神经血管耦合提供新的计算工具，有助于深入理解麻醉药物作用机理和量化效应，为研究麻醉意识消失机制提供新的理论基础。

项目在“麻醉的神经脑氧分析方法及临床应用”领域做了深入的研究。在基金的支持下，在国际专业领域期刊发表SCI论文11篇。该课题的进行为研究麻醉的药物作用机制提供了新颖的计算工具，有助于深入理解麻醉药物导致的意识消失及恢复，为提高麻醉的临床脑状态监测提供了技术和理论基础。本项目的主要成果包括：在麻醉下的EEG信号分析方面，提出了针对单通道的Renyi排序熵、多尺度样本熵、排序Lempel-Ziv复杂度等方法，以及基于双通道分析的排序交叉互信息方法。并详细对比了熵方法和同步方法在麻醉状态监测重点性能。在麻醉的模型机制研究方面，提出了从药物代谢到神经效应的机制模型及高性能的计算实现，建立了从丙泊酚和神经活动间的关系模型，为理解丙泊酚效应机制提供了模型工具。在实际的脑功能系统设计方面，设计了可用于临床的NIRS系统和EEG-NIRS系统，进行了初步的临床有效性评估。设计了全带宽的EEG系统和多通道神经电生理系统，可以用于高带宽、高采样率的脑电及神经元活动信号获取与分析。并结合Android平台，设计了一款便携式EEG监测系统。项目的研究成果对于推进麻醉的脑状态监测具有重要意义。