Pan-Tompkins演算法 : 濾波器設計

Abstract

中文摘要

在大部分QRS偵測方法中，每種方法基本上都能偵測到QRS的片段，像是[3]所提出的以微分為基礎的QRS偵測，此偵測方式能偵測到QRS發生的位置，但是卻因為演算法的關係無法及時偵測，以及在peak searching計算較為複雜，所以[4]提出了一樣以微分為基礎的演算法，但是peak searching 卻比[3]還要簡單，不過卻也無法及時偵測；因此為了做到即時偵測，Pan 及 Tompkins 提出了他們自己的演算法，藉由濾波器的設計以及不同的偵測方式，達到了及時的偵測。 此篇報告著重於此演算法的濾波器設計，因為在[1]中的濾波器皆是以取樣頻率為200Hz做設計，因此在取樣頻率改變的情況下，濾波器的轉移函數該如何調整，為本報告的重點內容。在這篇報告中，在驗證部分使用的是MATLAB (2018b, MathWorks®, Inc., USA)。

Abstract

Basically, most of QRS-detection method can detected QRS segment well, for example, reference [3] proposed a detection method based on derivative, this method can detected QRS when it was appeared in ECG signal. However, this algorithm do not detect QRS in an ongoing ECG signal, and the computation complexity of peak searching was not easy. To deal with the problem, reference [4] proposed the same algorithm, which was also based on derivative, to reduce the computation complexity of peak searching. The computation complexity of peak searching proposed by [4] is easier than peak searching in [3]. However, like [3], [4] do not detect QRS in an ongoing ECG signal. To approach real-time detection, Pan and Tompkins [1] proposed their QRS detection algorithm called Pan-Tompkins algorithm. This algorithm consisted of filtering and fiducial mark. This algorithm could detected QRS in an ongoing ECG signal well, and resolved the above problem well. This report focuses on the filter design of this algorithm. In [1], the filter sampling period is 1/200 seconds, if we changed the sampling period, the transfer function of filter need adjusting. In this report, I will modify the transfer function to deal with that problem. In this report, all of computer experiments were conducted in MATLAB (2018b, MathWorks®, Inc., USA).