RU2410023C1 - Method for qrs-complex electrocardiosignal extraction - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: continuous electrocardiosignal is recorded, filtered and presented as discrete records. In all derivations, the discrete records are summed over modulus. A threshold level is formed to be compared to a value of the sum over modulus of the discrete records of electrocardiosignal. A cardicycle, QRS-complex electrocardiosignal and R-R interval are extracted. In the extracted cardiocycle, values of derivatives Y' of the discrete records of electrocardiosignal are calculated. It is followed with extracting an interval of the values of an electrocardiosignal derivative belonging to the QRS-complex electrocardiosignal with the beginning and termination of the extracted interval of the values of the electrocardiosignal derivative are the beginning and termination of the QRS-complex electrocardiosignal.

EFFECT: more precise extraction of the beginning and termination of the QRS-complex.

3 cl, 11 dwg

Description

The invention relates to medicine, in particular to electrocardiology, and can be used to highlight the QRS complex in the automatic analysis of information parameters of an electrocardiogram (EX), performed by both hardware and software. Recognition and correct interpretation of the complex is a QRS-key point in assessing the activity of cardiomyocytes of the ventricles of the heart.

Currently, the most common methods for detecting the QRS complex of EX-based on the allocation of the R-R interval.

A known method for determining the QRS complex, implemented in a device for highlighting the QRS complex [1], which consists in the fact that the ECS is recorded, filtered, a threshold level is formed from the received signal and compared. According to the results of the comparison, a “time window” is formed, which allows to exclude false selections of the QRS complex. This method has the following disadvantages:

- reliable selection of QRS-complexes can be achieved only with reliable selection of the first (starting) QRS-complex;

- the effect of low-frequency additive interference on the ECS (polarization of electrodes, respiratory waves, artifacts, etc.) reduces the reliability of the allocation of the QRS complex; an attempt to remove the low-frequency additive interference using a high-pass filter will not give a result, since the cutoff frequency of such a filter should be low (0.05 Hz in accordance with the standards for electrocardiographs), and the action of artifacts leads to a transient process that lasts longer , the lower the filter cut-off frequency f, during the transition process, the reliability of QRS complex isolation decreases.

Closest to the proposed method (prototype) is a method of isolating the beginning of a cardiocycle [2], which consists in the fact that the ECS are recorded, filtered and presented in the form of discrete samples, a threshold level is formed with which the values of each discrete sample of the cardiac signal are compared, RR is isolated interval and cardiocycle. Then form a "time window", allowing to highlight the QRS-complex.

The figure 1 shows a diagram of the algorithm of the known method for highlighting the QRS complex of the cardiac signal.

The figure 2 shows a graph of an ideal electrocardiogram.

The figure 3 shows graphs of real electrocardiograms in 12 standard leads.

From the analysis of figure 1 and the description of the known method, it follows that the selection of the QRS complex is carried out by analyzing the EX in the amplitude-time domain.

Indeed, all the actions of the known method are carried out only in the amplitude-time domain. So, as a result of “comparing the values of each discrete readout of the electrocardiogram” with the threshold level, the R-R interval is allocated. Then, the cardiocycle is extracted and the “time window” is formed, with the help of which the QRS complex is extracted in the selected cardiocycle.

Analysis and comparison of images of ideal and real electrocardiograms (see figures 2 and 3) shows that it is difficult to reliably identify the boundaries of the QRS complex by analyzing the real EX in the amplitude-time domain. In EX, the beginning of the QRS complex is the end of the PQ segment, and the end of the QRS complex (point J) is the beginning of the ST segment. If on the ideal EX, the beginning and end of the QRS complex are clearly fixed (see figure 2), then the real EX is characterized by interference, as well as due to changes in the state of the heart at the beginning and end of the QRS complex, the occurrence of transient processes during registration, making it difficult to clearly distinguish QRS complex (see figure 3). In other words, in the amplitude-time domain in real EX there are no clear boundaries between the individual elements of the EX, which are subject to recognition.

From figure 1 it follows that the known method is based on a temporary analysis of the contour sections of the ideal electrocardiogram: PQ segment - the segment from the end of the P wave to the beginning of the Q wave, ST segment - the segment from the end of the QRS complex to the beginning of the T wave and the TP interval, the electrocardiogram signal between the teeth T and R. So, the author of the known method notes that "the most stable section of the electrocardiogram is the part of the isoline between the T and P teeth." Moreover, “the duration of the TP segment is significantly longer than the duration of the PQ segment and the ST segment. This fact is the basis of the method. " The figure 2 shows the image of the ideal electrocardiogram, which has the parameters necessary for the proper functioning of the known method, and which is similar to the image of the electrocardiogram shown in figure 2 of the prototype.

Thus, for the implementation of the known method, it is necessary that the PQ segment, the ST segment and the interval TP of the electrocardiogram signal are on the isoline. The analysis of real ECS given in the electrocardiographic literature [3, 4, 5, 6] shows that this condition is practically not fulfilled: a deviation from the isoline is always observed.

According to the authors of the present invention, the analysis of the information parameters of the ECS only in the amplitude-time domain is not enough to reliably identify the beginning and end of the QRS complex. It is necessary to expand the scope of presentation of the analyzed information. It is known that the integrated use of various approaches contributes to the solution of the problem of increasing noise immunity as the main element for achieving information integrity. Therefore, reliable identification of the boundaries of QRS complexes can be achieved only using various methods of analysis of EX.

The disadvantage of this method of allocating a QRS complex is, according to the authors of the present invention, the lack of reliability of the allocation of the beginning and end of the QRS complex.

The invention is aimed at increasing the reliability of highlighting the beginning and end of the QRS complex by analyzing the information parameters of the EX in the amplitude-phase region.

This is achieved by the fact that in the method of isolating the beginning of the cardiocycle, which consists in the fact that a continuous electrocardiogram is recorded, filtered and presented as discrete samples, the discrete samples are summed modulo in all leads, a threshold level is formed with which the sum value is compared modulo of discrete samples of an electrocardiogram signal, a cardiocycle and a QRS-comtex are isolated, an RR interval is additionally allocated, the values of the derivative Y 'of discrete samples of an electroc are calculated in the selected cardiocycle of a cardiosignal, allocate the range of values of the derivative of the electrocardiogram belonging to the QRS complex of the electrocardiogram, the beginning and end of the selected interval of values of the derivative of the electrocardiogram are the beginning and end of the QRS complex of the electrocardiogram.

Moreover, the allocation of the RR interval is carried out by additional bandpass filtering of the electrocardiogram, calculating the derivative F 'of the discrete samples of the filtered electrocardiogram according to the formula F = [F ₂ -F ₁ ; F ₃ -F ₂ ; ...; F _m -F _m-1 ], where F is a vector containing the values of discrete samples of the electrocardiogram, the value of the sum modulo in all leads which exceeds the threshold level, F 'is a vector containing the values of the derivative of discrete samples an electrocardiosignal, the value of the modulo of which exceeds the threshold level, m is the number of discrete samples in the filtered electrocardiogram, the sum of the modulo of which exceeds the threshold level, determine the vector L containing the numbers of discrete samples of the filtered elec a cardiofeedback signal, in which the value of the derivative F _i ^' (i = 1 ... m-1) is equal to zero, while the interval between adjacent values of the vector L is the RR interval of the cardiac signal.

In this case, the allocation of the interval of values of the derivative of the electrocardiogram belonging to the QRS complex of the electrocardiogram is carried out by calculating the derivative of the electrocardiogram according to the formula Y '= [Y ₂ -Y ₁ ; Y ₃ -Y ₂ ; ...; Y _n -Y _n-1 ], where Y is the vector containing the values of discrete samples of the electrocardiogram in the selected cardiocycle, Y 'is the vector containing the values of discrete samples of the derivative of the electrocardiogram in the selected cardiocycle, n is the number of discrete counts in the selected cardiocycle, the construction of the attractor of the electrocardiogram in the selected cardiocycle according to the formula

, где D - вектор расстояний между значениями дискретных отсчетов аттрактора электрокардиосигнала и началом координат, вычисления расстояний между значениями соседних дискретных отсчетов аттрактора по формуле

A = [{Y ₁ ^' , Y ₂ }; {Y ₂ ^' , Y ₃ }; ...; {Y ^' _n-1 , Y _n }], where A is the two-dimensional coordinate vector of the attractor, calculating the distances between the values of the discrete samples of the attractor of the cardiac signal and the origin using the formula

where D is the vector of distances between the values of the discrete samples of the attractor of the electrocardiogram and the origin, the calculation of the distances between the values of the neighboring discrete samples of the attractor by the formula

where P is the vector of distances between the values of adjacent discrete samples of the attractor, the calculation of the characteristics of discrete samples of the electrocardiogram located on the left in the sequence from the reference number l in the selected cardiocycle, which belongs to the vector L, according to the formula

where S1 is the vector of characteristics of the discrete samples located to the left of the sample l, k = 1 ... l-1, i = 1 ... n-1, the calculation of the characteristics of the discrete samples of the electrocardiogram located to the right in order from the reference number l in the selected cardiac cycle, which belongs to the vector L, by the formula

, где QRS_start - дискретный отсчет, соответствующий началу QRS-комплекса электрокардиосигнала, k=1…l-1, определения правой границы интервала дискретных отсчетов, принадлежащих QRS-комплексу электрокардиосигнала, по формуле

, где QRS_stop - дискретный отсчет, соответствующий окончанию QRS-комплекса электрокардиосигнала, k=1…(n-1)-(l-1).where S2 is the vector of characteristics of the discrete samples located to the right of the point l, k = 1 ... (n-1) - (l-1), i = 1 ... n-1, the definition of the left border of the interval of discrete samples belonging to the QRS-complex of the cardiac signal , according to the formula

where QRS _start is a discrete sample corresponding to the beginning of the QRS complex of an electrocardiogram, k = 1 ... l-1, determining the right boundary of the interval of discrete samples belonging to the QRS complex of an electrocardiogram, according to the formula

where QRS _stop is a discrete sample corresponding to the end of the QRS complex of an electrocardiogram, k = 1 ... (n-1) - (l-1).

The figure 4 shows a diagram of the algorithm of the proposed method for allocating a QRS complex.

The figure 5 shows a diagram of the algorithm for allocating the interval of values of the derivative of the electrocardiogram belonging to the QRS complex of the electrocardiogram.

The figure 6 shows the image of the attractor real electrocardiogram.

The figure 7 shows the image of the attractor of the real electrocardiogram with marked points belonging to the QRS complex of the electrocardiogram.

The figure 8 shows the image of the attractor of the real electrocardiogram in the form of individual values of discrete samples, not connected by a common line.

и

.The figure 9 shows the rationale for the signs

and

.

The figure 10 shows the selection of the QRS-complex electrocardiogram.

The essence of the proposed method is as follows. A continuous electrocardiogram is recorded, filtered and presented in the form of discrete samples. A threshold level is formed with which discrete samples of the electrocardiogram are compared. Then R-R intervals and cardiocycles are distinguished. In the selected cardiac cycle, processing of the ECS in the phase space is carried out, the coordinates of which are the amplitude and time derivative of the electrocardiogram. To do this, a graph is plotted on which the values of discrete samples of the vector Y are plotted along the abscissa axis, and the values of discrete samples of the vector Y 'are plotted along the ordinate axis. The resulting graph is a phase self-portrait or attractor EX (see figures 6 and 7). Attractor in English means “attractor”. It is known that the cardiovascular system (CVS) has a strange attractor due to the fractal nature of the functioning of the CVS. The attractor is an invariant of the individual state of the CVS and has a characteristic individual form in the phase space [7, 8]. Attractor properties are determined only by the internal psychophysiological characteristics of a given person.

The construction and study of the attractor is the most reliable and accurate method for isolating the portion of the cardiosignal corresponding to the QRS complex. The well-known method of highlighting the QRS complex is based on the time frame for establishing the boundaries of the QRS complex, and not on the parameters of the signal itself. At the same time, the time limits of the QRS complex are determined on the basis of statistical data and the features and form of the individual’s electrocardiogram are not taken into account.

Using the construction of the attractor of various cardiac signals, it was established that the attractor section corresponding to the QRS complex is a group of discrete samples, the values of which differ significantly from the values of other discrete samples of the attractor. This is because in the area of the electrocardiogram corresponding to the QRS complex, the intensity of the signal change (respectively, the processes in the heart) is high. Regardless of the amplitude of the values of the QRS complex on the electrocardiogram on the attractor plot, the portion of the QRS complex is always pronounced compared to other sections of the electrocardiogram and represents a “loop” of significant diameter (see figures 6 and 7). Another distinctive feature of the attractor section corresponding to the QRS complex of the electrocardiogram is a large interval between the values of adjacent discrete samples of the attractor (see figure 8).

, где k=1…l-1, окончания интервала признаку

, где k=1…(n-1)-(l-1), где S1, S2 - векторы характеристик дискретных отсчетов слева и справа от отсчета l на электрокардиосигнале соответственно.The analysis of figures 6, 7 and 8 shows that the criterion for determining the interval of values of the derivative of the electrocardiogram belonging to the QRS complex of the electrocardiogram is the correspondence of the beginning of the interval to the sign

where k = 1 ... l-1, the end of the interval

, where k = 1 ... (n-1) - (l-1), where S1, S2 are the characteristics vectors of discrete samples to the left and to the right of the reference l on the cardiac signal, respectively.

The characteristics of the discrete samples are calculated based on the above features of the attractor section corresponding to the QRS complex: a “loop” of significant diameter with the largest distances between adjacent values of the discrete samples (see Fig. 8). The discrete reference characteristic is a combination of three components:

1) the distance from the value of the discrete reference on the attractor to the origin,

2) the distance from the value of the studied discrete reference to the value of the next discrete reference on the attractor,

3) the distance (remoteness) of the discrete count from the peak of the R wave on the cardiac signal.

Components 2) and 3) are of little importance compared to component 1), therefore, when taking them into account in the aggregate characteristic, it is necessary to apply coefficients. For component 2), the coefficient is equal to the ratio of the maximum distance from the value of the discrete sample on the attractor to the origin to the maximum distance between adjacent discrete samples on the attractor. For component 3), the coefficient is equal to the ratio of the maximum distance from the value of the discrete sample on the attractor to the origin to the number of discrete samples in the studied cardiocycle. The use of multiplying the components by the coefficients balances their influence in the aggregate characteristic, allows you to take into account all the components equally.

Thus, to distinguish the interval of the QRS complex of the cardiac signal, two characteristic vectors are compiled: the first for discrete samples located to the left of the peak of the R wave, according to the formula

where S1 is the vector of characteristics of discrete samples located to the left of the discrete sample l, k = 1 ... l-1, i = 1 ... n-1, the second for discrete samples located to the right of the peak of the R wave, according to the formula

where S2 is the vector of characteristics of discrete samples located to the right of the point l, k = 1 ... (n-1) - (l-1), i = 1 ... n-1. The peak of the R wave is determined by a discrete count l. The figure 9 shows the graphs of the characteristics of discrete samples in comparison with the original electrocardiograms.

где k=1…l-1, QRS_start - дискретный отсчет, соответствующий началу QRS-комплекса электрокардиосигнала,

, где QRS_stop - дискретный отсчет, соответствующий окончанию QRS-комплекса электрокардиосигнала, k=1…(n-1)-(l-1). На фигуре 9 видно, что при сопоставлении графика характеристик с изображением исходного электрокардиосигнала началу и окончанию QRS-комплекса на электрокардиосигнале соответствуют дискретные отсчеты с минимальными значениями характеристик на графике характеристик. Эта особенность используется для выделения начала и окончания QRS-комплекса электрокардиосигнала.The criteria are used to determine the boundaries of the interval of the QRS complex of an electrocardiogram

where k = 1 ... l-1, QRS _start is a discrete sample corresponding to the beginning of the QRS complex of an electrocardiogram,

where QRS _stop is a discrete sample corresponding to the end of the QRS complex of an electrocardiogram, k = 1 ... (n-1) - (l-1). Figure 9 shows that when comparing the characteristics graph with the image of the original electrocardiogram, the beginning and end of the QRS complex on the electrocardiogram correspond to discrete samples with the minimum values of the characteristics on the characteristics graph. This feature is used to highlight the beginning and end of a QRS complex of an electrocardiogram.

The figure 10 shows graphs of practical electrocardiograms with allocated to them using the proposed method QRS-complexes of electrocardiogram.

CONCLUSION

Thus, the proposed method allows to eliminate these disadvantages of the known method and to ensure reliable selection of the QRS complex of the cardiac signal, regardless of the following factors:

- possible deviations from the norm of the parameters of the teeth of the electrocardiogram: shape, amplitude, duration;

- drift of the EX isoline contour under the influence of additive low-frequency interference (the effect of respiration, artifacts, temporal drift, etc.),

which helps to improve the conditions for the subsequent processing of the electrocardiogram (calculation of the time parameters of the individual elements of the electrocardiogram, calculation of the duration of the cardiocycle, etc.).

In addition, a positive additional effect of the proposed method is a more reliable allocation of the point J of the electrocardiogram. Point J has important diagnostic value: it delimits the processes of de- and repolarization. Usually, the location of point J is considered to be a point located at a certain distance either from the beginning of the QRS complex, for example, by 80 or 60 ms [9], or from the top of the R wave by 40 ms [10]. At the same time, another point belonging to the tooth T is also set, which relative to the point J serves to determine the trend of the S-T segment.

An analysis of the known methods shows that in them the location of the point J is determined on the basis of the use of a priori known ratios of information parameters of the ideal EX in the studied EX. This approach leads to a predetermined probabilistic nature in determining the location of point J. That is, the location of point J is not carried out by analyzing the information parameters of the registered EX, but by “superimposing” the “time window” template on the registered EX. As a result, the specific value of the location of the point J is determined indirectly, which, of course, reduces the accuracy of the known methods. The determination of the location of point J should be carried out directly on the basis of the analysis of the registered EX, in particular, the QRS complex and the ST interval. According to the authors of the present invention, a probabilistic approach in determining the location of point J should be used to clarify the position of point J.

The technical and economic effect of the proposed method is to increase the reliability of highlighting the beginning of the QRS complex.

Information sources

1. RF patent 2021752, A61B 5/0452. A device for highlighting QRS complexes. / B.I. Kruk, N.I. Belkin // BI 1994, 20.

2. RF patent 2195164, АВВ 5/02. A method for isolating the beginning of a cardiocycle and a device for its implementation. / A.A. Mikheev // 12/27/2002.

3. Zudbinov Yu.I. ECG alphabet. - Rostov n / a: Phoenix, 2003, 240 p.

4. Kechker M.I. Guidelines for Clinical Electrocardiography. - M .: 2000, 395 p.

5. Murazhko VV, Strutinsky A.V. Electrocardiography // Tutorial. Ed. 6th - M.: MEDpressinform, 2004, 320 p.

6. Orlov V.N. Guide to electrocardiography. - M .: Medicine, 1984, 528 p.

7. Anischenko B.C., Yanson N.B., Pavlov A.N. Can a healthy person's heart regimen be regular? // Radio engineering and electronics, 1997, v. 42, No. 8, 1005-1010 p.

8. Vishnevsky V.V., Ragulskaya M.V., Fainzilberg L.S. The effect of solar activity on the morphological parameters of the ECG of the heart of a healthy person. // Journal of Radio Electronics (electronic journal of the RAS), 2002, No. 12.

9. Ryabykina G.V., Sobolev A.V. Heart Rate Variability: Monograph. - M.: Star'ko, 1998, 200 p.

10. Zaichenko K.V., Zharinov O.O., Kulin A.N., Kulygina L.A., Orlov A.P. The removal and processing of bioelectric signals: a Training manual. / Ed. K.V. Zaichenko. SPbGUAP, SPb., 2001, 140 pp.

Claims (3)

1. A method for extracting a QRS complex of an electrocardiosignal, namely, that a continuous electrocardiogram is recorded, filtered and presented as discrete samples, the discrete samples are summed modulo in all leads, a threshold level is formed with which the sum value is compared modulo discrete samples of the electrocardiogram, the cardiocycle and the QRS complex of the electrocardiogram are distinguished, characterized in that the RR interval is distinguished, the values of the derivative Y 'are calculated in the selected cardiocycle electrocardiosignal samples, isolated interval values derivative electrocardiosignal belonging QRS-complex electrocardiosignal, the beginning and ending values of the selected interval electrocardiosignal derivative are the beginning and end of QRS-complex electrocardiosignal. 2. The method according to claim 1, characterized in that the allocation of the RR interval is carried out by additional bandpass filtering of the electrocardiogram, calculating the derivative F 'of the discrete samples of the filtered electrocardiogram according to the formula F' = [F ₂ -F ₁ ; F ₃ -F ₂ ; ...; F _m -F _m-1 ], where F is a vector containing the values of the discrete samples of the filtered electrocardiogram, the value of the sum modulo in all leads which exceeds the threshold level, F 'is a vector containing the values of the derivative of the discrete samples of the filtered electrocardiogram, the value of the sum modulo which exceeds the threshold level, m is the number of discrete samples in the filtered electrocardiogram, the value of the modulo sum of which exceeds the threshold level, the definition of the vector L containing the numbers of discrete samples s of the filtered electrocardiogram, in which the value of the derivative F _i ^' (i = 1 ... m-1) is zero, and the interval between adjacent values of the vector L is the RR interval of the electrocardiogram. 3. The method according to claim 2, characterized in that the allocation of the interval of values of the derivative of the electrocardiogram belonging to the QRS complex of the electrocardiogram is carried out by calculating the derivative of the electrocardiogram according to the formula Y '= [Y ₂ -Y ₁ ; Y ₃ -Y ₂ ; ...; Y _n -Y _n-1 ], where Y is the vector containing the values of the discrete samples of the electrocardiogram in the selected cardiocycle, Y 'is the vector containing the values of the discrete samples of the derivative of the electrocardiogram in the selected cardiocycle, n is the number of discrete counts in the selected cardiocycle, the construction of the electrocardiogram attractor in the selected cardiocycle according to the formula A = [{Y ₁ ^' , Y ₂ }; {Y ₂ ^' , Y ₃ }; ...; {Y _n-1 ^' , Y _n }], where A is the two-dimensional coordinate vector of the attractor, calculating the distances between the values of the discrete readings of the attractor of the electrocardiosignal and the beginning of ordinates of the formula

,
where D is the vector of distances between the values of the discrete samples of the attractor of the electrocardiogram and the origin, the calculation of the distances between the values of the neighboring discrete samples of the attractor by the formula

where P is the vector of distances between the values of adjacent discrete samples of the attractor, the calculation of the characteristics of discrete samples of the electrocardiogram located to the left in the order of the discrete sample with number l in the selected cardiocycle, which belongs to the vector L, by the formula

where S1 is the vector of characteristics of the discrete samples located to the left of the discrete sample with the number l, k = 1 ... l-1, i = 1 ... n-1, the calculation of the characteristics of the discrete samples of the electrocardiogram located to the right in order from the discrete sample with the number l in the selected cardiocycle, which belongs to the vector L, according to the formula

where S2 is the vector of characteristics of discrete samples located to the right of the discrete report with number l, k = 1 ... (n-1) - (l-1), i = 1 ... n-1, determining the left border of the interval of discrete samples belonging to QRS -complex of an electrocardiogram, according to the formula

where QRS _start is a discrete sample corresponding to the beginning of the QRS complex of an electrocardiogram, k = 1 ... l-1, determining the right boundary of the interval of discrete samples belonging to the QRS complex of an electrocardiogram, according to the formula

where QRS _stop is a discrete sample corresponding to the end of the QRS complex of an electrocardiogram, k = 1 ... (n-1) - (l-1).

Images