使用自相关和时间序列检测的方法检测心室纤颤信号

心室纤颤信号必须准确而及时地识别,如果不使用除颤器及时进行除颤,病人会迅速死亡。另一方面,如果心室纤颤信号没有发生,而误判为心室纤颤信号,给予电击,则病人的心脏受到不应有的损伤,这也属于严重的医疗事故。因此快速与准确的识别显得十分重要。自70年代以来,世界上许多科学家都在探讨各种方法,以期望使用计算机自动识别心室纤颤信号,但均因达不到实时判别的要求或准确率不够高而不能用于自动除颤器检测的设计方案。本文介绍了一种准确率高而又较简易的快速识别法。这一方法是取一秒心电信号,进行有限长度离散信号的自相关,然后对该自相关函数作零切割,让自相关函数变成一串二进制脉冲串,接着以一秒信号的脉冲串数作为特征参数,设计分类器。经证明,心室纤颤信号(VF信号)(Ventricular Fibrillation)和心动过速信号(VT信号)(Ventricular Tachycar-dia)的特征参数均属于高斯分布。因此可用Wald时间序列检测方法对VF信号和VT信号分类。这个算法可用于计算机进行实时处理和分析,同时此法与取样率无关,适用于各种取样系统,是一个较理想的方法。

Ventricular Fibrillation Detection by Autocorrelation and Sequential Testing

Ventricular Fibrillation (VF)should be immediately and accura-tely detected. Patients would die ina few minutes if VF is not abolis-hed with a defibrillator. It is veryimportant to detect VF accuratelyand timely. Since 70s scientistsin the world have succeded in seek-ing some methods for automatic de-tection of VF signals by using com-puter, but some of them are suit-able for designing automatic defibri-llator for not being able to reachthe standard of accuracy and time-liness. In this paper we develop arapid recognition algorithm withvery high accuracy. This method is1) Take one second ECG sig-nal, compute the autocorrelation.2) Convert the the ECG sig-nal into a binary sequence by comp-arison with a zero-crossing.3) Differentiate VF by testingimpulse sequence numbers. The dis-tribution of the impulse numberson autocorrelation of one secondVT signals and VF signals are de-monstrated to be Gaussian. We canuse Wald sequential detection fordifferentiation VF from VT. Thismethod is simpler and faster thanever pre-existing methods. It is sui-table for implementation in a real-time microcomputer system. Thisalgorithm does not depend on sam-pling and is suitable for implemen-tation of different sampling system