LeBlanc A R
Crit Rev Biomed Eng. 1986;14(1):1-43. doi: 10.1080/08913818608459473.
Quantitative analysis of cardiac arrhythmias has been the subject of intensive research during the last 10 years. Several systems have been designed to help in the processing of cardiac signals: single or multiple lead electrocardiograms (ECG), electrograms from intracardiac catheters, esophageal recordings, etc. The main objective of these developments was oriented toward positive identification of arrhythmias or rhythm-disturbance counts to imitate the cardiologist's interpretation in contexts such as routing ECG and ambulatory recordings. However, these systems were mainly measurement tools aimed at extracting auricular and ventricular depolarization timings plus gross morphology description. The domain of morphology analysis of beat-to-beat auricular depolarization on ECG has never been highly active due to poor signal conditions. For routine ECG, automatic interpretation was set as an objective to complement computer-assisted ECG interpretation of conduction problems (i.e., morphology analysis of a representative beat extracted or averaged from the dominant rhythm). The limitations of rhythm interpretation in this context are well known. In the ambulatory ECG context, the analysis procedures are relatively simple and are often summarized as trivial counts describing, most exclusively, the ventricular arrhythmic behavior of the heart over a relatively long duration. Waveform-detection and measurement have been the bottleneck of advancement in arrhythmia analysis since highly reliable detection of events on a beat-to-beat basis are necessary to perform a valid analysis. Rare approaches have proposed probabilistic definition of event detection. The present review puts emphasis on the potential of several methods which have been demonstrated as powerful in identifying short- or long-duration heartbeat patterns, mode of heartbeat initiation, mode of heartbeat coupling, etc. Globally, these methods are referred to as time series analysis, modeling of rhythm patterns, simulation, and pattern recognition. A delay in the advancement of the study of arrhythmogenesis and limiting the analysis of arrhythmias to textbook descriptions is not justified when put in perspective of the potential of implementing powerful techniques which have been more or less neglected or used in a narrow way.
在过去十年中,心律失常的定量分析一直是深入研究的主题。已经设计了几种系统来辅助处理心脏信号:单导联或多导联心电图(ECG)、心内导管记录的电信号图、食管记录等。这些进展的主要目标是朝着心律失常的阳性识别或节律紊乱计数发展,以模仿心脏病专家在诸如常规心电图和动态记录等情况下的解读。然而,这些系统主要是测量工具,旨在提取心房和心室去极化时间以及总体形态描述。由于信号条件不佳,心电图上逐搏心房去极化的形态分析领域一直不太活跃。对于常规心电图,自动解读被设定为一个目标,以补充计算机辅助的心电图传导问题解读(即从主导节律中提取或平均得到的代表性搏动的形态分析)。在这种情况下,节律解读的局限性是众所周知的。在动态心电图情况下,分析程序相对简单,通常总结为简单的计数,主要是描述心脏在相对较长时间内的室性心律失常行为。波形检测和测量一直是心律失常分析进展的瓶颈,因为要进行有效的分析,必须在逐搏基础上对事件进行高度可靠的检测。很少有方法提出事件检测的概率定义。本综述强调了几种方法的潜力,这些方法已被证明在识别短期或长期心跳模式、心跳起始模式、心跳耦合模式等方面非常强大。总体而言,这些方法被称为时间序列分析、节律模式建模、模拟和模式识别。从实施强大技术的潜力来看,心律失常发生机制研究的进展延迟以及将心律失常分析局限于教科书描述是不合理的,这些技术或多或少被忽视或使用方式有限。