## A Supervised Wavelet Transform Algorithm for R Spike Detection in Noisy ECGs ⋆

### BibTeX

@MISC{Lannoy_asupervised,

author = {G. De Lannoy and A. De Decker and M. Verleysen},

title = {A Supervised Wavelet Transform Algorithm for R Spike Detection in Noisy ECGs ⋆},

year = {}

}

### OpenURL

### Abstract

Abstract. The wavelet transform is a widely used pre-filtering step for subsequent R spike detection by thresholding of the coefficients. The time-frequency decomposition is indeed a powerful tool to analyze non-stationary signals. Still, current methods use consecutive wavelet scales in an a priori restricted range and may therefore lack adaptativity. This paper introduces a supervised learning algorithm which learns the optimal scales for each dataset using the annotations provided by physicians on a small training set. For each record, this method allows a specific set of non consecutive scales to be selected, based on the record’s characteristics. The selected scales are then used for the decomposition of the original long-term ECG signal recording and a hard thresholding rule is applied on the derivative of the wavelet coefficients to label the R spikes. This algorithm has been tested on the MIT-BIH arrhythmia database and obtains an average sensitivity rate of 99.7 % and average positive predictivity rate of 99.7%. 1

### Citations

1836 | Robust statistics - Huber - 1981 |

54 |
Detection of ECG characteristic points using wavelet transforms
- Li, Zheng, et al.
- 1995
(Show Context)
Citation Context ...ause it allows precise time-frequency representation of the signal with a low computational complexity. A lot of work has been published in past years on the use of the WT for QRS detection. In 1995, =-=[20]-=- used an algorithm based on finding the maxima larger than a threshold obtained from the pre-processed initial beats. Later, [17] produced a method allocating a R peak at a point being the local maxim... |

38 |
The principles of software QRS detection
- Kohler, Henning, et al.
- 2002
(Show Context)
Citation Context ...thods have been reported previously: template matching [12], mathematical models [24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods =-=[19]-=-, hidden Markov models [10], neural networks [31,27], adaptative thresholding [16,9] and geometrical approach [28]. In all cases, using a pre-filtering step prior to beat detection is required for the... |

19 |
ECG Beat Detection Using Filter Banks
- Afonso, Tompkins
- 1999
(Show Context)
Citation Context ...logical variations due to the patient and its disease make the ECG a non-stationary signal. (2) Other ECG components such as the P or T wave looking like QRS complexes often lead to wrong detections. =-=(3)-=- There are many sources of noise that pollute the ECG signal such as power line interferences, muscular artifacts, poor electrode contacts and baseline wanderings due to respiration. These three probl... |

18 | time electrocardiogram QRS detection using combined adaptive threshold
- Christov, “Real
(Show Context)
Citation Context ...24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding =-=[16,9]-=- and geometrical approach [28]. In all cases, using a pre-filtering step prior to beat detection is required for the elimination of artifacts in the signal. The time-frequency decompositions by wavele... |

18 |
Comparing Wavelet Transforms for Recognizing Cardiac Patterns
- Senhadji, Carrault, et al.
- 1995
(Show Context)
Citation Context .... A lot of work has been published in the literature about the detection of various interesting ECG features such as P waves, QRS waves, T waves, QT intervals or abnormal beats by numerous techniques =-=[2,25,26]-=-. This paper focuses on R spike detection only. For this purpose, several approaches using different signal processing methods have been reported previously: template matching [12], mathematical model... |

13 |
Software QRS detection in ambulatory monitoring – a review, Med Biol Eng Comput
- Pahlm, S¨ornmo
- 1984
(Show Context)
Citation Context ...This paper focuses on R spike detection only. For this purpose, several approaches using different signal processing methods have been reported previously: template matching [12], mathematical models =-=[24]-=-, signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding [16... |

9 |
Wavelet transforms and the ECG: A review
- Addison
- 2005
(Show Context)
Citation Context .... A lot of work has been published in the literature about the detection of various interesting ECG features such as P waves, QRS waves, T waves, QT intervals or abnormal beats by numerous techniques =-=[2,25,26]-=-. This paper focuses on R spike detection only. For this purpose, several approaches using different signal processing methods have been reported previously: template matching [12], mathematical model... |

9 |
PhysioNet: New Research Resource for Complex Physiologic Signals, Circulation 101(23): 2000
- Goldberger
(Show Context)
Citation Context ...CG. 5 Results and Validation The learning of the model on the training set and the assessment of performances must be done using an annotated database. The public standard MIT-BIH arrhythmia database =-=[14]-=- was used in this work. It contains 48 half-hour recordings of annotated ECG with a sampling rate of 360Hz and 11-bit resolution over a 10-mV range. The recorded signals contain different wave types a... |

9 |
Automated ECG segmentation with dynamic time warping
- Vullings, Verhaegen, et al.
- 1998
(Show Context)
Citation Context ...ent signal processing methods have been reported previously: template matching [12], mathematical models [24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping =-=[30]-=-, syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding [16,9] and geometrical approach [28]. In all cases, using a pre-filtering step prior to beat detec... |

9 |
Neural-network-based adaptive matched filtering for QRS detection
- Xue, Hu, et al.
- 1992
(Show Context)
Citation Context ...ching [12], mathematical models [24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks =-=[31,27]-=-, adaptative thresholding [16,9] and geometrical approach [28]. In all cases, using a pre-filtering step prior to beat detection is required for the elimination of artifacts in the signal. The time-fr... |

8 |
Using wavelet transform and fuzzy neural network for VPC detection from the holter ECG
- Shyu, Wu, et al.
(Show Context)
Citation Context ...ching [12], mathematical models [24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks =-=[31,27]-=-, adaptative thresholding [16,9] and geometrical approach [28]. In all cases, using a pre-filtering step prior to beat detection is required for the elimination of artifacts in the signal. The time-fr... |

7 |
Heart rate variability: a review
- Acharya, Joseph, et al.
- 2006
(Show Context)
Citation Context ...e variability (HRV) analysis [29]. HRV measures have been proven successful in diagnosing cardiac abnormalities and neuropathies or evaluating the actions of the autonomic nervous system on the heart =-=[1]-=-. However, HRV measures heavily rely on the accuracy of the QRS feature detection on the digitalized ECG signal. Automatic feature extraction and especially R spike detection is thus a milestone for E... |

7 |
F.: Wavelet transform-based QRS complex detector
- Kadambe, Murray, et al.
- 1999
(Show Context)
Citation Context ...blished in past years on the use of the WT for QRS detection. In 1995, [20] used an algorithm based on finding the maxima larger than a threshold obtained from the pre-processed initial beats. Later, =-=[17]-=- produced a method allocating a R peak at a point being the local maxima of several consecutive dyadic wavelet scales. In both these methods, a postprocessing allowed to eliminate false R detections. ... |

7 |
A Wavelet Tour of Signal Processing: Wavelet Analysis & Its Applications
- Mallat
- 1999
(Show Context)
Citation Context ...x, in order to emphasizes these complexes and to filter the useless noise. For more details on the wavelet transform and on the standard wavelet functions available, the interested reader can consult =-=[21,2,11]-=-. 4 Methodology 4.1 General Description The detection of R spikes is a tough task due to the complexity of the ECG signal. The aim of the algorithm introduced here is to automatically find the best su... |

7 |
A wavelet-based ECG delineator: evaluation on standard databases
- Martίnez, Almeida, et al.
- 2004
(Show Context)
Citation Context ...both these methods, a postprocessing allowed to eliminate false R detections. Based on these two publications, a lot of other researches were published on the beat detection using a WT filtering step =-=[27,13,22,2,8,7]-=-. The main problem of the WT is that one has to choose the mother wavelet and the scales used to analyze the signal on an empirical basis. While the mother wavelet can easily be chosen based on its ch... |

7 |
Delineation of the QRS complex using the envelope of the e.c.g
- Nygards, Sornmo
- 1983
(Show Context)
Citation Context ... R spike detection only. For this purpose, several approaches using different signal processing methods have been reported previously: template matching [12], mathematical models [24], signal envelop =-=[23]-=-, matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding [16,9] and geometrical a... |

4 |
J.C Cornily " Automatic P-wave analysis of patients prone to atrial fibrillation
- Clavier, Lepage
- 2002
(Show Context)
Citation Context ...eviously: template matching [12], mathematical models [24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models =-=[10]-=-, neural networks [31,27], adaptative thresholding [16,9] and geometrical approach [28]. In all cases, using a pre-filtering step prior to beat detection is required for the elimination of artifacts i... |

3 |
An algorithm for computer measurement of QT intervals in the 24 h ECG
- Algra, Zeelenberg
- 1987
(Show Context)
Citation Context ...al approaches using different signal processing methods have been reported previously: template matching [12], mathematical models [24], signal envelop [23], matched filters [18], ECG slope criterion =-=[4]-=-, dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding [16,9] and geometrical approach [28]. In all cases, using a pre-filteri... |

3 |
A real-time QRS detection method based on moving-averaging incorporating with wavelet denoising
- Chen, Chen, et al.
- 2006
(Show Context)
Citation Context ...both these methods, a postprocessing allowed to eliminate false R detections. Based on these two publications, a lot of other researches were published on the beat detection using a WT filtering step =-=[27,13,22,2,8,7]-=-. The main problem of the WT is that one has to choose the mother wavelet and the scales used to analyze the signal on an empirical basis. While the mother wavelet can easily be chosen based on its ch... |

3 |
A novel approach in R peak detection using Hybrid Complex Wavelet,‖ Int
- Fard, Moradi, et al.
(Show Context)
Citation Context ...both these methods, a postprocessing allowed to eliminate false R detections. Based on these two publications, a lot of other researches were published on the beat detection using a WT filtering step =-=[27,13,22,2,8,7]-=-. The main problem of the WT is that one has to choose the mother wavelet and the scales used to analyze the signal on an empirical basis. While the mother wavelet can easily be chosen based on its ch... |

2 |
Wavelet methods for spike detection in mouse renal sympathetic nerve activity
- Brychta, Tuntrakool, et al.
- 2007
(Show Context)
Citation Context ...lar morphology of the QRS complex. The wavelet is represented Fig. 1. It is defined as the second derivative of the gaussian probability density function: ψ(t) = 1 √ 2πσ 3 ( 1 − t2 σ 2 ) e −t2 2σ 2 . =-=(5)-=-260 G. de Lannoy, A. de Decker, and M. Verleysen Fig. 1. The Mexican hat mother wavelet In order to select the appropriate scales among the wide range of wavelet scales, one needs a criterion. A natu... |

2 |
Dsp implementation of wavelet transform for real time ecg waveforms detection and heart rate analysis. Computer Methods and program in Biomedicine
- Chen, Chen, et al.
- 1997
(Show Context)
Citation Context ...rror rate is below one percent. To our knowledge, only the R spike detectors based on Li’s algorithm obtained comparable results with a sensitivity and a positive predictivity between 99.7% and 99.9% =-=[22,20,6]-=-. However, this algorithm makes use of several heuristic rules and requires the setting of many empirical parameters. Our algorithm achieves comparable performances with a very simple non-parametric t... |

2 |
Detection of qrs-wave in electrocardiogram: Based on kalman and wavelets
- Chen, Yan, et al.
- 2005
(Show Context)
Citation Context |

2 |
Qrs detection by template matching using real-time correlation on a microcomputer
- Dobbs, Schmitt, et al.
- 1984
(Show Context)
Citation Context ...ous techniques [2,25,26]. This paper focuses on R spike detection only. For this purpose, several approaches using different signal processing methods have been reported previously: template matching =-=[12]-=-, mathematical models [24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], a... |

2 | A new approach to qrs segmentation based on wavelet bases and adaptive threshold technique. Medical Engineering and
- Madeiro, Cortez, et al.
- 2007
(Show Context)
Citation Context ...24], signal envelop [23], matched filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding =-=[16,9]-=- and geometrical approach [28]. In all cases, using a pre-filtering step prior to beat detection is required for the elimination of artifacts in the signal. The time-frequency decompositions by wavele... |

2 |
An automated approach to beat-by-beat qt-interval analysis
- Sahambi, Tandon, et al.
- 2000
(Show Context)
Citation Context .... A lot of work has been published in the literature about the detection of various interesting ECG features such as P waves, QRS waves, T waves, QT intervals or abnormal beats by numerous techniques =-=[2,25,26]-=-. This paper focuses on R spike detection only. For this purpose, several approaches using different signal processing methods have been reported previously: template matching [12], mathematical model... |

2 |
Ecg beat detection using a geometrical matching approach
- Silva, Berthoumieu, et al.
- 2007
(Show Context)
Citation Context ...d filters [18], ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding [16,9] and geometrical approach =-=[28]-=-. In all cases, using a pre-filtering step prior to beat detection is required for the elimination of artifacts in the signal. The time-frequency decompositions by wavelet transform258 G. de Lannoy, ... |

1 |
den Akker, T.J.: Beat-to-beat interval measurement in the electrocardiogram
- Koeleman, Ros, et al.
- 1985
(Show Context)
Citation Context ...y. For this purpose, several approaches using different signal processing methods have been reported previously: template matching [12], mathematical models [24], signal envelop [23], matched filters =-=[18]-=-, ECG slope criterion [4], dynamic time warping [30], syntactic methods [19], hidden Markov models [10], neural networks [31,27], adaptative thresholding [16,9] and geometrical approach [28]. In all c... |