基于混合神经网络的ECG数据压缩 |
| |
引用本文: | 张浙亮,吕维雪. 基于混合神经网络的ECG数据压缩[J]. 浙江大学学报(工学版), 2000, 34(1): 71-76 |
| |
作者姓名: | 张浙亮 吕维雪 |
| |
作者单位: | 东方通信技术中心!浙江杭州310013(张浙亮),浙江大学生命科学与医学工程学系!浙江杭州310027(吕维雪) |
| |
摘 要: | 提出了一种基于神经网络自适应模板匹配(NNATM)的ECG压缩方法,利用Kohonen的自组织特征映射(SOFM)对ECG波形进行分类,然后根据分类结果将心搏波形送往相应的前馈网络压缩,并生成一匹配模板,最后用SAPA3算法对残差信号进行处理。对以250Hz和11bit量化的MIT/BIH心电波型的实验表明,该方法压缩效果好于SAPA3算法。
|
关 键 词: | 心电数据 数据压缩 神经网络 模板匹配 心脏疾病 |
修稿时间: | 1996-11-21 |
ECG data compression using neural-network-based adaptive template matching method |
| |
Abstract: | A new ECG data compression method using neural|network|based adaptive template matching (NNATM) method is presented. To compress ECG, we used a Kohonen's self|organizing feature map (SOFM) to classify waveforms, and sent them to correspondent multilayer feedforward neural networks to compress according to class label to generate a proper ECG template; Finally, we used SAPA3 algorithm to process the residual signals. The suggested method was evaluated by using ten kinds of MIT/BIH ECG waveforms which are sampled with 250 Hz and 11 bit/sample. The results showed that this method improved the accuracy of the reproduced signals while the compression ration was higher comparing to SAPA3 algorithm. |
| |
Keywords: | electrocardiogram (ECG) data compression neural network template matching |
本文献已被 CNKI 维普 等数据库收录! |