胸外按压优化决策与抗按压干扰除颤节律辨识算法研究

Chest compression (CC) and defibrillation have been the cornerstones of cardiopulmonary resuscitation (CPR). At present, there are two problems needed to be solved during CC, which are the lack of an effective trade-off between blood flow Improvement and ribs fracture reduction and the disability to detect shock rhythm from CPR artifact corrupted ECG using the ECG alone. On the basis of summarizing previous works, this study will be carried out as the three major parts: extraction of the characteristic parameters, algorithm building and experimental tests. Firstly, based on animal model, the key characteristic parameters and data which can identify the degree of blood flow, chest injury sensitively and the time-frequency characteristic of CC artifact and original ECG signal will be extracted. Secondly, the rule for identifying the degree of blood flow and ribs fracture will be designed; the original ECG waveform will be estimated; the algorithm for optimal CC strategy and rhythm identification during CC will be finally established. Thirdly, the support identification system and the support decision system for CPR will be developed with the designing idea of virtual medical instrument, and will be compared with the traditional CPR treatment; the algorithm will then be improved for the clinical application. The algorithms described by the present study may be taken into account as a resolution for intelligent and automatic CPR. Therefore, with the present algorithm, it is likely to promote the popularized application of CPR and improve the survival rate of cardiac patient and the success rate of CPR.

胸外按压和电击除颤是抢救心脏骤停患者的核心急救措施。目前，在该领域亟待解决的问题主要表现在以下两个方面：1）缺乏权衡血流灌注与骨折风险的胸外按压决策优化能力；2）缺乏无需额外硬件支持的抗按压干扰除颤节律辨识能力。本项目拟在前期研究工作基础上，以数据挖掘、建立算法、实验评估为主线，分三个阶段开展研究。首先，基于动物模型，挖掘血流灌注程度和骨折风险的特征参数，以及原始胸外按压分量和原始心电信号分量的时频域特征值；其次，设计血流灌注、骨折风险分级辨识规则，重建原始心电波形，建立胸外按压优化策略和抗按压干扰心电节律辨识算法；最后，研制辨识决策虚拟仪器系统，开展对比实验，对辨识决策效果做出整体评价；基于评价结果，进行算法优化，实现临床移植。本项目有望解决心肺复苏智能化和自动化过程中的瓶颈问题，为心肺复苏技术的普及推广，提高心肺复苏的成功率和心脏骤停患者的存活率，提供必要的理论和技术支持。

项目背景：胸外按压和电击除颤是抢救心脏骤停患者的核心急救措施。目前，在该领域亟待解决的问题主要表现在以下两个方面：（1）缺乏权衡血流灌注与骨折风险的胸外按压决策优化能力；（2）缺乏无需额外硬件支持的抗按压干扰除颤节律辨识能力。.主要研究内容：本项目以数据挖掘、建立算法、实验评估为主线，分三个阶段开展研究。首先，基于动物模型，挖掘血流灌注程度和骨折风险的特征参数，以及原始胸外按压分量和原始心电信号分量的时频域特征值；其次，设计血流灌注、骨折风险分级辨识规则，重建原始心电信号波形，建立胸外按压优化策略和抗按压干扰心电节律辨识算法；最后研制辨识决策虚拟仪器系统，开展对比试验，对辨识决策效果作出整体评价；基于评价结果，进行算法优化，实现临床移植。.重要结果：1. 攻克了权衡血流灌注程度与胸骨骨折风险的胸外按压优化决策算法；2. 攻克了无需额外硬件支持的抗按压干扰除颤节律辨识算法；3.发表论文9篇，其中SCI/EI索引论文6篇；4. 申请发明专利2项；5. 成功研制胸外按压优化决策与抗按压干扰除颤节律辨识系统原理验证样机1套；6. 培养博士研究生1名，硕士研究生4名。.关键数据：1.相比传统按压，本研究实现的优化按压获得了更大的权衡优化评分。优化按压较传统按压取得了更好的血流灌注。9组数据中有6组的有益评分值大于传统按压。另一方面，优化按压获得了比传统按压更为安全的胸外按压，将骨折风险有效保持在了最高风险等级以下。2. 基于本课题建立的抗按压干扰除颤节律辨识算法获得了优秀的辨识效果：在SNR=-12dB的情况下辨识结果的敏感性为92.1%，特异性为86.6%，准确率为88.6%。准确率比未滤波算法高37%，比传统固定截止频率滤波器高24.5%。.科学意义：本项目有望解决心肺复苏智能化和自动化过程中的瓶颈问题，为心肺复苏技术的普及推广，提高心肺复苏的成功率和心脏骤停患者的存活率，提供必要的理论和基础支持。