心电图房颤信号提取的主分量分析研究

Atrial fibrillation signals extraction from ECG based on principal component analysis

为从体表心电图估计房颤期间患者的心房活动状态,需进行房颤信号提取.基于主分量分析研究,提出了一种从房颤患者的单导联体表心电图中提取房颤信号的新方法.由于Holter记录通常不能提供足够的空间信息,通过利用连续心博中心房和心室活动存在时间相关的特征,应用主分量分析去相关的思想从单导联进行QRST波消除方法.通过对Physionet数据库中的大量房颤数据测试和验证,表明该方法克服了固定模板平均心博抽取技术对波形形态变化的敏感性,使被估计信号的QRST残余新信号最小,获得了一种更鲁棒、效果更好的单导联房颤波提取技术.

For estimating the atrial activity of patients with atrial fibrillation （AF）, AF signals must be subtracted. In Holter records, the few number of leads was not enough to exploit the spatial information of electrocardiology （ECG）. The proposed approach uncorrelated the time dependence of ventricular and atrial components in continuous cardiac beats by principal component analysis（PCA）. The methodology was tested and validated using a significative database with AF recordings from Physiobank. The method＇s main benefit compared to average beats template subtraction is its insensibility to variations of the QRST morphology because of reasons such as breathing etc, and so minimizing the residual QRST. The method provides a more robust extraction technology of AF signals from single-lead ECG records.