【作者】 王英华；

【导师】 李小俚；

【作者基本信息】 燕山大学 ， 模式识别与智能系统， 2012， 硕士

【摘要】 麻醉深度监护对于指导麻醉用药，减少手术风险和病人痛苦具有重要意义。传统监护方法主要基于病人的自主反应和心率变化、自发性表皮肌电等生理参数，靠麻醉师的经验来估计，缺乏清晰的量化指标。近年来，基于头皮脑电信号(Electroencephalogram, EEG)的麻醉深度监测技术得到了广泛的重视，并有多款EEG麻醉深度监护产品出现。然而这些监护产品主要基于线性系统理论，分析非线性的EEG信号存在缺陷；而且价格昂贵，难以在国内推广。因此，探讨新的麻醉深度监测算法，并据此研制有独立技术的麻醉深度监护仪具有重要的现实意义。论文提出了一种基于数字信号处理器(Digital signal processor, DSP)的麻醉深度监护系统方案。分析了影响脑电信号分析的各种噪声的来源、特点和常见去噪方案，提出了适合麻醉深度监护和DSP计算的去噪算法。探讨了现有的脑电信号分析方法，并对基于排序熵的麻醉深度监护算法进行了详细介绍，包括算法原理、药代药效动力学分析、统计分析等。通过与其他参数的比较显示了排序熵的优越性能。文中还探讨了一种基于多尺度排序熵的麻醉深度监测算法。通过药代药效动力学分析和与另外几种常见麻醉监护参数的比较，表明该方法也能够很好地反映麻醉深度造成的脑电变化，并在一定程度上揭示了脑电信号的多尺度特性。论文介绍了包括前端放大器的设计、通道切换电路、数据采集逻辑、数据处理、数据传输及上位接收显示的完整硬件实现。在此基础上，介绍了DSP平台上的程序设计和核心的算法实现。包括数据采集、USB通信的基本原理和实现、相关滤波算法和排序熵算法的实现等。同时讨论了程序设计和算法优化中的一些关键问题。更多还原

【Abstract】 Monitoring of the depth of anesthesia (DoA) is of great importance in guiding theapplication of anesthetics, avoiding sugery risks and reducing patients’pains. TraditionalDoA monitoring are generally based on patient’s reactions and physiological parameterssuch as heartrate variation and spontaneous skin myoelectricity. The judgement of DoAhighly relies on the experience of the anesthetist, and lacks a defined quantative measure.Great attention has been paid to DoA monitoring via electroencephalogram (EEG) analy-sis recently. A few products have emerged on the market. However, most products calcu-late their paramaters based on linear theory, therefore have deficits in analyzing non-linearEEG signals. The high price also restricts their application in China. Therefore, deveploingnew DoA monitoring algorithms and devices has a great significance.A DoA monitor based on digital signal processor (DSP) is proposed in this disserta-tion. are discused. The source, characteristics and removal of artifacts within EEG are de-scribed, and de-noise solutions suitable for DoA monitoring and DSP realization are pro-posed. Some available EEG analysis methods are presented, and the DoA monitoring al-gorithm based on Permutation Entropy (PE) is described in detail, including the principleof PE, the pharmacokinetics/pharmacodynamics (PK/PD) analysis and statistical analysis.The comparision of PE with other indexes proved its advantage. Moreover, a DoA moni-toring algorithm based on the multiscale sample entropy is proposed. PK/PD analysisshows its validity in reflecting EEG changes with anesthetic concentration. The method isproved better than a few DoA paramaters. Paticularly, it reveals the multiscale characteris-tic of EEG signal to some extend.The complete design of an EEG acquistition system is explained in this dissertation,including the front amplifier, channel muxing, acquistition logic, data processing, datatransmission and the data receive and display on the host system. Then the implement ofthe key algorithms on the DSP platform are explained. Main code for data acquistition,USB communication, digital filters and PE calculation are explained. Issues on program-ming and algorithm optimization are discussed.更多还原