A Non-parametric Derivative-based Method For R Wave Detection

ECG signal is one of the earliest biological signals which is studied and appliedto clinical medicine by human. Its regularity can be directly observed through the useof electrocardiogram. Accurate automatic ECG analysis and diagnosis play a key rolein cardiovascular diseases. QRS wave is the most visible part of the changes in theECG (electrocardiogram) and the synthetic performance of the multi-myocardial cells.However QRS wave is difficultly detected, not only is there the physiologicalvariability of the QRS wave but also several various types of noises that wouldpresent in ECG signals. The main several noises sources include muscle noises,artifacts due to electrode motion,50Hz power frequency interference and baselinewander. So detecting QRS wave accurately is not only provide important basis for thediagnosis of arrhythmia, but also it is possible to count the heart rate and thevariability of heart rate in this base.In this paper, a detection statistics method of R waves based on non-parameter isbeing proposed. This method firstly uses a digital filter to cut out noises from ECGsignals, utilizes local polynomial estimation that is a non-parametric derivative-basedmethod to estimate original data the derivative values, and then selects appropriatethresholds by the difference, and the algorithm automatically adjusts the size ofthresholds periodically according to the different needs. Afterwards, the position of Rwave is detected by the estimation of the first-order derivative values. In order toimprove the accuracy of detection, the method of redundant detection and missingdetection is applied in this paper. In addition, we also use local polynomial estimationto fit the RR interval which is obtained from the detected R wave and discuss its heartrate variability. The MIT-BIH ECG database and clinical data Massachusetts Instituteof Technology are used for experimental testing and the experimental results showthat the method is simple, easy to implement, high accuracy. This statistics method ofR waves based on non-parameter is more efficient and accurately than differencethreshold method. It has good flatness on the role of ECG data. ECG data via localpolynomial fitting is more accurately reflect the fluctuation of raw ECG. The detectionof non parameter statistics method is applied to the ECG signal is the main innovationof this paper.