动态心电信号数据压缩的神经网络特征算法

Neural network algorithm for data compression of signals of ambulatory ECG

目的：建立基于神经网络的心电数据压缩算法。方法：根据ＥＣＧ特征的变化，从ＥＣＧ中提取出对诊断有意义的信息。并采用了反向传播算法（ＢＰ算法）的三层感知机，构造了两个相同结构的神经网络。结果：本系统的压缩精度在２％～５％左右；压缩比在１０～２０左右。结论：这样构成的压缩系统能适应心电信号的变化，且具有较高的压缩比和压缩精度。

Objective: To propose a new algorithm based on neural network for data compression of signals of ambulatory ECG. Methods: The neural network was used to select those ECG signals valuable for diagnosis for it was a 3-layered perception with back propagation algorithm and the compression system was composed of two neural networks with the same structures. Results: The compression rate was about 10～20 times and its precision about 2%～5%. Conclusion: This system have high compression rate and precision.