一种基于神经网络的心电图数据压缩方法

A data compression approach of ECG using neural network

本文提出了一种新的基于ＢＰ神经网络的心电图（ＥＣＧ）数据压缩方法，其要点包括：①根据心电图特点进行分段压缩，②采用改进的ＢＰ算法；③压缩过程中自适应调整隐单元的状态值。该方法具有重建误差小、适应性强、压缩比高的特点。本文对它的压缩性能进行了计算机模拟研究，结果表明，在Ｈｏｌｔｅｒ系统中应用该方法进行心电图数据压缩是切实可行的。

A new data compressing aproach of electrocardiogram (ECG) using neural network is presented.Its main components include: 1.Segment compression according to feature of ECG; 2. Improved BP algorithm; 3. Adaptive adjustinent of hidden units'activation levels in compression. This new approach features strong adaptation and high compression natio. Computer simulation results of the performance study show that it is feasible to implement ECG data compression using this new approach for Holter system.