ECG fiducial points extraction by extended Kalman filtering

Abstract

Abstract-Most of the clinically useful information in Electrocardiogram (ECG) signal can be obtained from the intervals, amplitudes and wave shapes (morphologies). The automatic detection of ECG waves is important to cardiac disease diagnosis. In this paper, we propose an efficient method for extraction of characteristic points of ECG. The method is based on a nonlinear dynamic model, previously introduced for generation of synthetic ECG signals. For estimating the parameters of model, we use an Extendend Kalman Filter (EKF). By introducing a simple AR model for each of the dynamic parameters of Gaussian functions in model and considering separate states for ECG waves, the new EKF structure was constructed. Quantitative and qualitative evaluations of the proposed method have been done on Physionet QT database (QTDB). This method is also compared with a method based on Partially Collapsed Gibbs Sampler (PCGS). Results show that the proposed method can detect fiducial points of ECG precisely and mean of estimation error of all FPs (except Ton) do not exceed five samples (20 msec).

Advanced P Wave Detection in Ecg Signals During Pathology: Evaluation in Different Arrhythmia Contexts

“…For a more thorough comparison, proposed algorithm as well as various well-known methods 15,18,24,26 were tested on the manually annotated part of the QTDB. The methods 15 and 24 are based on dyadic WT (wavelet quadratic spline) and PT, respectively.…”

“…This area is demarcated with respect to RR interval and QRS complex fiducial points (onset and offset). Once the area has been defined, the maximum of P wave is generally detected directly, e.g., by adaptive threshold estimation 12,13 , using wavelet transform (WT) 14–16 , through extraction of P wave template and the application of correlation 17 , Kalman filtering 18 , moving average 13 , support vector machine 19,20 , Prony’s method 21 , the hidden Markov models 22 , a neural network 23 , phasor transform (PT) 24,25 , dynamic programing 26 , or combination of several detection algorithms 27 . These approaches deliver good results only in the cases of ECGs with normal cardiac rhythm, but these approaches were not tested on pathological records.…”

ECG fiducial point extraction using switching Kalman filter

“…In [21,22,23,24], methods based on EKF have been proposed. The main limitation of such methods is their sensitivity to the initial location of the Gaussian functions as 35 well as initial parameters of EKF, that must be defined by the user.…”

“…We also have a comparison with our previously proposed methods (linear and nonlinear EKF25 [23,24]). Validation and comparison are done over Physionet QT database [44,45].…”

“…In Table 1, we compare the proposed SKF method with other methods of the literature: Wavelet [17], PCGS [15] and EKFbased method (EKF17) [22]. In Table 2, we compare SKF with our previously proposed 175 methods based on EKF: linear and nonlinear EKF25 methods proposed in [23] and [24]. Most of the records of QT database do not have reference label for T on .…”