Automatic Detection of ECG Ventricular Activity Waves using Continuous Spline Wavelet Transform

Abstract

--In this study we present detection algorithms of characteristic points of the QRS and T waves based on the continuous wavelet transform (CWT) with splines. This technique can handle any integer scale and the analysis is not restricted to scales that are powers of two, which allows to use a wide range of scales and to more efficiently reduce noise and artifacts. Evaluation of the QRS detection algorithm performance has been done in eight ECG data files of the MIT-BIH database, and the accuracy has been of 99.5 %. Evaluation of the detection algorithms of the QRS wave onset and offsets of QRS and T waves has been done in the CSE multi-lead measurement database, and the measurements were within the tolerance limits for deviations with respect of the manual measurements determined by the CSE experts. Therefore, the proposed algorithms to detect characteristic points of the QRS and T waves based on this technique allow the evaluation of the CWT in more scales, are robust to noise and artifacts and have the accuracy of a human expert.

An algorithm for beat-to-beat heart rate detection from the BCG based on the continuous spline wavelet transform

“…The algorithm for J p detection based on the CWT with splines above described has been tested in seven longitudinal BCG records (six lasting for 100 s and one lasting for 60 s) on seated subjects. R-R intervals from the simultaneous recorded ECG were obtained with an algorithm for QRS detection based on the CWT with splines using the scale 2, which has an accuracy of 99.5% [22]. To quantitatively evaluate the algorithm, the agreement between R-R intervals from ECG and J-J intervals from BCG and their corresponding heart rates were assessed by using the statistical method proposed by Bland and Altman [25].…”

“…where the notation ([p]↑ m × u n 2 m )(k) represents the kth term of the convolution of the sequences p upsampled by a factor m and of sequence u n 2 m , the filter u n 2 m is a cascade of (n 2 + 1) moving sum filters of order (m − 1) with an offset k 0 that ensures its symmetry, b n 1 +n 2 +1 is the B-spline representation of a spline of order n 1 + n 2 + 1, and c(k)'s are the B-splines coefficients. The program developed by Alvarado et al [22] (written in Matlab ® , The MathWorks Inc.), calculates the CWT of the discrete signal x(t) at the integer scales m with expansion coefficients spline p, and its implementation is based in the fast algorithm proposed by Unser et al [21]. It can be seen from (6) that filters are iterated discrete convolutions of moving sums, and can be computed without any multiplication, which results in a very efficient algorithm.…”

Algorithms Based on CWT and Classifiers to Control Cardiac Alterations and Stress Using an ECG and a SCR

“…This method is not influenced by the motion of the user so the analysis can be done when the user is practicing sport. In [25] the authors use CWT with a success of the 99.5%.…”

“…Alvarado et al [25] propose a method based on the CWT with splines taking various scales into consideration to detect the QRS complex, obtaining an accuracy of 99.5%.…”

A new fitting approach for online electrocardiogram component waves delineation

“…Timing characterization of the P-and T-waves in the ECG was developed in [8] using wavelet transform with a modified mother wavelet. Alvardo et al [9] used different algorithms based on the continuous wavelet transform with splines to delineate the QRS and T-waves. In [10] an optimized version of the MOBD algorithm [11] was applied to delineate the QRS-complex, whereas a different method based on the Joeng algorithm [12] was used for the P-and T-waves delineation.…”