论文部分内容阅读
致死性心电节律的辨识和分类是自动体外除颤仪的关键任务。本文对已存在的心电节律辨识算法提取出的21个特征值进行了回顾性研究,并基于这些特征值构建了一个遗传算法优化的反向传播神经网络。以数据库提供的1 343例心电信号样本用于实验。实验结果表明,本文构建的神经网络在对窦性节律、心室颤动、室性心动过速、心脏停搏4类心电信号的辨识分类上有很好的表现,在测试集上的平衡准确性高达99.06%;相较已存在的算法,辨识性能更好。将该算法应用在自动体外除颤仪上,将进一步提高除颤前节律分析的可靠性,最终提高心脏骤停的存活率。
The identification and classification of lethal ECG rhythm is a key task of an automated external defibrillator. In this paper, 21 eigenvalues extracted from existing rhythm identification algorithms are retrospectively studied. Based on these eigenvalues, a genetic algorithm optimized back propagation neural network is constructed. A sample of 1 343 ECG samples provided by the database was used for the experiment. The experimental results show that the neural network constructed in this paper has good performance in the classification and classification of four kinds of ECG signals of sinus rhythm, ventricular fibrillation, ventricular tachycardia and cardiac arrest. The accuracy of the balance in the test set Up to 99.06%. Compared with existing algorithms, the recognition performance is better. Applying this algorithm to an automated external defibrillator will further improve the reliability of rhythm analysis prior to defibrillation and ultimately improve the survival rate of cardiac arrest.