摘要
在远程心电诊断中 ,计算机辅助ECG诊断通常是在接收到ECG信号、进而解压重建后进行的 ,这样便造成诊断工作的延误。本研究提出了一种基于AR模型的ECG直接分类方法 ,利用AR模型系数及其建模误差作为特征对ECG信号进行压缩 ,并采用非线性二次判别函数形式进行特征分类。通过对MIT BIH标准数据库中的NSR、APC、PVC、SVT、VT和VF各 2 0 0个样本信号进行测试 ,获得了 93.5 %~ 97.86 %的分类精度。该方法的特点是 :诊断迅速方便 ,能同时对多类ECG信号进行有效分类 。
In telecardiogram diagnosis systems, remote Electrocardiogram (ECG) feature extraction and diagnosis are usually performed after the reconstruction of ECG signals, so computer-assisted automatic diagnosis would be delayed due to the reconstruction and feature extraction. A technique based on autoregressive(AR) modeling was proposed, AR coefficients and modeling errors were used to accomplish the compression of ECGs and then classify ECGs by using discriminant of nonlinear quadratic function. The experiments by using MIT-BIH standard data demonstrated that the accuracy of classification for arrhythmias, such as NSR, PVC, APC, VT, VF, and SVT, is from 93.5% to 97.86%. The Characeristic of this method is quick diagnosing speed and versatile recognition efficiency, and therefore it is especially suitable for remote telecardiogram diagnosis application.
出处
《中国生物医学工程学报》
EI
CAS
CSCD
北大核心
2004年第3期222-229,共8页
Chinese Journal of Biomedical Engineering
关键词
远程心电
ECG信号
AR建模
二次判别函数
特征提取
Computer aided diagnosis
Computer simulation
Electrocardiography
Feature extraction
Mathematical models