RU2251968C1 - Method and device for eliminating electrocardiogram signal isoline drift - Google Patents

Abstract

FIELD: medicine; medical engineering.

SUBSTANCE: method involves selecting reference point in every cardiac cycle on TP-segment. Values of neighboring N=2n+1 reference points also belonging to TP-segment are recorded, n=1,2,…, beginning from the first reference point. Other reference points are set to zero. The central reference point value is left without changes in a group of 2n+1 member. Reference point values of each of n pairs of reference points symmetrically arranged relative to the central reference point are scaled relative to condition U_j=U₀K_j, where U₀ is the central reference point amplitude, U_j is amplitude of j-th reference point pair, j=1,2,…,n is the number of each reference point pair relative to the central reference point, K_j is the scaling coefficients determined from received signal suppression condition of the first n spectral zones in spectrum. The so formed electrocardiogram signal reference point groups sequence is let pass through lower frequency filter with isoline drift signal being obtained being produced on output. The signal is amplified and subtracted from the initial electrocardiogram signal that is preliminarily delayed for lower frequency filter delay time. Device has the first lower frequency filter, discretization unit and unit for selecting anchor reference points connected in series, as well as subtraction unit, unit for saving N reference points, scaling unit, the second lower frequency filter, amplifier and delay unit. Output of the unit for selecting anchor reference points is connected to the first input of memory unit the second input of which is connected to discretization unit output. Each of N memory unit outputs is connected to one of N inputs of scaling units. Scaling unit output is connected to the second lower frequency filter input which output is connected to amplifier input. Amplifier output is connected to the first input of subtraction unit, the second output of subtraction unit is connected to delay unit output. Its input is connected to output of the first lower frequency filter. Subtraction unit output is the device output.

EFFECT: reliable removal of isoline drift.

2 cl, 8 dwg

Description

The invention relates to medicine, in particular to electrocardiography, and can be used in the processing of electrocardiograms (EX).

The contour drift can be due to the action of additive low-frequency interference on the EX (polarization of the electrodes, the effect of respiration, artifacts, time drift, etc.).

Currently, the most common methods for eliminating isoline drift are based on filtering low-frequency noise and on interpolation reconstruction of the isoline drift signal with its subsequent subtraction from the initial EX.

Elimination of low-frequency noise using high-pass filters leads to distortion of the low-amplitude elements of the EX, for example, the ST-segment. In addition, the action of artifacts leads to a transient process that lasts longer, the lower the filter cutoff frequency, i.e. the contour drift occurs again.

A known method of compensating the drift of the contour [1], which consists in the fact that in each cardiocycle a point is taken that is taken as an isoelectric. The amplitude parameters of the studied EX elements (R-wave, ST-segment, etc.) are defined as the difference between the measured amplitude at a particular point and the amplitude of the point taken as isoelectric.

This method has the following disadvantage. Complete compensation of the contour drift is possible only when this drift is a constant within one cardiocycle displacement above or below the actual contour. Typically, the drift of the isoline of a real EX is in the form of a sinusoid (respiratory waves) or an exponent (transient), therefore, it cannot be fully compensated by the described method.

A known method of compensating the drift of the contour [2], taking into account the change in the signal drift of the contour within the cardiocycle. According to this method, ECS are sampled, reference points (samples) located on the PQ segment are extracted in each cardiocycle, approximating functions are obtained from these points, including the angular coefficients of lines drawn between adjacent P-Q points. Next, the compensated values of y (i) EX, free from the action of the drift of the contour line, are found in accordance with the formula y (i) = x (i) -m-k (i-ipq), where i is the number of the discrete count of the EX; x (i) is the value of the i-th sample; m is the P-Q level of the point of the current cardiocycle; k is the slope of the line connecting neighboring P-Q points; (i-ipq) is the distance of the i-th sample from the corresponding P-Q point.

This method has the following disadvantages:

1) even with a sinus rhythm, the pronounced PQ segment lying on the isoline is not always present in the ECS, which makes it difficult to select reference points;

2) by the proposed method, it is possible to compensate for the drift of the isoline of only a very low frequency compared to the heart rate (HR), when it is fair to replace the value of the function with the value of its argument.

Closest to the proposed method (prototype) is a method of eliminating the drift of the contour [3], which consists in the fact that on the original EX, which is a mixture (sum) of the cardiac signal and low-frequency noise (drift of the contour), is isolated, as in the method [2 ], the reference points located on the PQ segment measure the values of the mixture of the EX and low-frequency noise (isoline drift) at these points, the parameters (coefficients) of the interpolating spline function are calculated from these values, and this function is constructed at all points of the processing yvaemogo pacemaker then subtracted value spline function of starting the pacemaker.

This method has the following disadvantages:

1) in the same way as in the method [2], the express PQ segment is not always present in the EX, the sequence of which is necessary for constructing the spline function;

2) when the contour drift frequency increases, the accuracy of recovering the contour drift signal deteriorates by the spline function, and when half the heart rate is reached, restoring the contour drift becomes impossible and, therefore, its elimination is impossible;

3) the delay of the restored isoline drift signal relative to the original EX signal does not fully compensate for the drift of the EX isoline contour.

The proposed method allows to eliminate these disadvantages.

The technical result of the invention consists in expanding the functionality of the method and device by reliably eliminating the drift of the contour at frequencies up to and even higher than the heart rate.

The analysis of electrocardiograms with various deviations from the norm showed that when removing the electrocardiogram in a clinic or hospital, that is, when the patient is at rest, the most stable section of the electrocardiogram is the part of the isoline between the T and P waves. Two more areas lying on the ECS [4] on the isoline: segment PQ - a segment from the end of tooth P to the beginning of tooth Q; ST segment - a segment from the end of the ORS complex to the beginning of the T wave. The PQ segment reflects the physiological delay in the transmission of the excitation pulse and can normally be isoelectric or slightly negative. The ST segment characterizes the period of complete coverage by excitation of the ventricles, therefore, an isoline is recorded on the electrocardiogram at this moment. Disorders in the myocardium displaces this segment above or below the contour. The TP segment corresponds to the electrical diastole of the heart. If a low-frequency additive interference, which manifests itself in the form of an isoline drift, acts on the EX, then only the signal of this interference is present on the TR segment.

The essence of the proposed method is as follows. The electrocardiosignal, cleared of high-frequency and industrial-frequency interference, which is the initial signal for subsequent processing, is sampled, a reference sample is allocated in each cardiac cycle on the TR EX section, and, starting from this reference, the values N = 2n + 1 samples are stored, n = 1, 2 , ... lying on the TP segment, the remaining samples within the cardiocycle are equal to zero, the central sample value of the thus formed group of (2n + 1) samples is left unchanged, and the sample values of each of the n pairs of samples located symmetrically with respect to the central reference, scale in accordance with the condition U _j = U ₀ K _j , where U ₀ is the amplitude of the central reference, U _j is the amplitude of the samples of the jth pair of samples, j = 1, 2, ..., n is the number each pair of samples relative to the central reference, To _j - scale factors, defined as a solution to the system of equations

where τ is the duration of the counting of the EX,

T is the repetition period of the reference samples.

Next, the obtained sequence of signals of the groups of samples is passed through a low-pass filter. The signal from the output of the low-pass filter of the drift is amplified, receiving the contour signal, this signal is subtracted from the original signal EX, which is previously delayed by the delay time of the low-pass filter.

The proposed method allows to eliminate the drift of the isoline of the pacemaker when changing the frequency of the latter from zero to frequencies comparable to and even higher than the heart rate, which improves the conditions for the subsequent processing of the cardiac signal (calculation of the time and amplitude parameters of individual elements of the cardiosignal).

Let us explain the principle of achieving a technical result by performing the above actions with an electrocardiogram. Any sample of the ECS mixture and low-frequency noise in the form of an isoline drift (Fig. 1, a), selected on the TP segment of the cardiac signal, is a noise reference, since the TP segment corresponds to the electric diastole of the heart. Thus, we obtain the amplitude-pulse modulation of the interference signal (Fig.1, b). Spectrum of an interference signal with amplitude-pulse modulation provided that the theorem of V.A. Kotelnikov is fulfilled, i.e. the sampling frequency Fo = 1 / T, where T is the sampling period, more than twice the maximum interference frequency Fп, shown in Fig 2, a. The spectrum contains interference components Fп at the frequency and spectral zones at the sample repetition frequencies Fo, 2Fo, 3Fo, etc. Each spectral zone contains lateral components kFo ± Fп, where k = 1, 2, ... is the number of the spectral zone. With increasing frequency of the isoline drift signal, the component at the interference frequency shifts in the spectrum to the right along the frequency axis, and the left side component of the first spectral zone shifts to the left. If the conditions of the theorem of V.A. Kotelnikov are violated, the frequency of the left side component becomes less than the component at the interference frequency (Fig. 2, b), and the isolation of the isoline drift signal by spline approximation methods becomes impossible.

We show how to isolate a low-frequency additive interference signal (isoline drift) from an ECS mixture and interference even when the interference frequency exceeds half the heart rate, which in this case determines the interference sampling frequency.

If we add a pair of samples to the sample of the interference signal located on the TR EX segment, one of which is adjacent to the original sample on the left and the other is symmetrical on the right, then in the spectrum of the new signal, each sample of which now consists of three pulses, any spectral zone can be suppressed under the condition

where k is the number of the suppressed spectral zone,

τ is the duration of the reference pulse and additional pulses,

T is the repetition period of the samples.

For practical reasons, it is advisable to set the value of τ convenient for implementation (usually this value is an integer number of times less than the sampling period T) and solve the equation for U ₁ .

If you want to suppress another spectral zone, for example, with the number m, then you need to add to the existing three pulses a couple more pulses, one of which is adjacent to the leftmost pulse and the other is adjacent to the rightmost pulse from the right. Each signal sample is now represented by a group of five samples. To suppress the spectral zones with numbers k and m in such a signal, it is necessary to solve the system of two equations

The value of the central reference U ₀ , which is the initial one, is known, therefore, the system is solved based on the above considerations regarding U ₁ and U ₂ .

In the general case, the number of equations in the system is equal to the number of suppressed spectral zones, for example, n, while the number of pulses representing each sample of the signal is 2n + 1, and the number of terms in each equation is n + 1. If the spectral zones with numbers 1, 2, ..., n are suppressed, then the system of equations will take the form

As in previous cases, the system is solved with respect to U _j , j = 1, 2, ..., n and denotes the number of the newly added pair of samples located symmetrically with respect to the central sample. In the spectrum of such a signal, n first spectral zones will be suppressed, but the zero spectral zone containing the interference component will be preserved (Fig. 2, c). The number n of suppressed spectral zones is selected so that the frequency of the left side component of the (n + 1) -th spectral zone is greater than the interference frequency. Under this condition, the component at the frequency of the interference can be extracted from the general signal using a low-pass filter. Since in real conditions of processing an electrocardiogram signal, the values of its samples can only be changed using the scaling operation, we introduce the concept of the scale factor K _j = U _j / U ₀ , and write the above system of equations in the form

This system of equations is solved with respect to K _j .

Figure 3 shows the structural diagram of a device for implementing the proposed method of eliminating the drift of the isoline of the cardiac signal, figure 4 is an embodiment of block 3, figure 5 is a comparison circuit 13, figure 6 and 7 are explanations of the operation of block 3, and on Fig - time diagrams explaining the operation of the device.

To achieve a technical result and implement the proposed method, a unit for storing values N = 2n + 1, n = 1 is introduced into a device containing a series-connected low-pass filter, the input of which is the input of the device, a sampling unit and a block for allocating reference samples, as well as a subtraction unit , 2, ..., samples, scaling unit, second low-pass filter, amplifier and delay unit. The output of the reference sample allocation unit is connected to the first input (control) of the storage unit, the second input of which (information) is connected to the output of the sampling unit, each of the N outputs of the storage unit is connected to one of the N inputs of the scaling unit, the output of the scaling unit is connected to the input of the second filter low-frequency, the output of which is connected to the input of the amplifier, the output of the amplifier is connected to the first input of the subtraction unit, the second input of the subtraction unit is connected to the output of the delay unit, the input of which is connected to the input of the unit while sampling, the output of the subtraction block is the output of the device.

A device for implementing the proposed method for eliminating the contour drift comprises a first low-pass filter 1, a sampling unit 2, a reference sample allocation unit 3, a memory unit 4, a scaling unit 5, a second low-pass filter 6, an amplifier 7, a subtraction unit 8, and a delay unit 9.

The input of the filter 1, which is the input of the device, receives an electrocardiogram. The output of the filter is connected to the input of the sampling unit 2 and to the input of the delay unit 9, the output of the sampling unit 2 is connected to the input of the allocation unit of the reference sample 3 and to the second input of the memory unit 4, the first input of which is connected to the output of the selection unit of the reference sample 3, each of N the outputs of the storage unit 4 is connected to one of the N inputs of the scaling unit 5, the output of the scaling side 5 is connected to the input of the second low-pass filter 6, the output of which is connected to the input of the amplifier 7, the output of the amplifier 7 is connected to the first input of the subtracting unit anija 8, the second input of the delay 9, the input of which is connected to the input of the sampling unit is connected to the subtracting unit 8 to the output unit, the output of the subtractor 8 is the output device.

The device operates as follows. The electrocardiose signal, cleaned by filter 1 of high-frequency and network noise, but containing low-frequency noise in the form of an isoline drift (shown by the solid line in FIG. modeled in amplitude in accordance with the type of electrocardiogram following with a sampling period T. This sequence of sample pulses is fed to the input of the reference sample selection block a. One of the possible embodiments of block 3 is shown in figure 4. It contains a block 10 for generating second-order differences, a source of positive 11 and negative 12 threshold levels, a comparison circuit 13, a clock generator 14, a first circuit And 15, a pulse counter 16 and a second circuit And 17.

From the output of the sampling unit 2, discrete samples of the electrocardiosignal (EX) are fed to the input of the second-order difference generator 10, which is the input of the block 3 allocation of reference samples. At the output of block 10, there are signals of second-order differences, formed from three consecutive ECS samples

ddU _i = U _i -2U _i-1 + U _i-2 ,

where i are the numbers of samples involved in the formation of the next second-order difference,

U is the amplitude of the corresponding reference.

The received signals of the second-order differences are compared using comparison circuit 13 with two threshold levels, one of which is positive + Up (output of block 11), and the second negative - Up (output of block 12). The comparison circuit 13 can be performed (figure 5) based on two comparators 18 and 19 and the OR circuit 20.

The operation of the block 3 allocation of reference samples, performed according to the schemes of Fig.7 and 4 is modeled using the program circuit simulation MicroCAP5. In the simulation, a 6-bit pulse counter was used. The simulation results are presented in Fig.6 and 7.

When the signal of the second order difference is between the threshold levels (Fig.6), the signal at the output of the comparison circuit 13 has a high level potential (Fig.7a), which is fed to the input R of the counter 16, allowing it to work in the counting mode, and one of the inputs of the first circuit And 15, allowing the passage of its output to the other input of the clock pulses from the generator 14. The repetition rate of the clock pulses is equal to the sampling frequency of the cardiac signal.

The clock pulses from the output of the circuit And 15 go to the counting input From the counter 16 (Fig.7, b), which carries out the count of these pulses. The corresponding bit outputs of the counter 16 (Fig.7, c, ..., h) are connected to the corresponding inputs of the second circuit And 17. The signal in the form of a rectangular pulse of positive polarity (Fig.7, and) appears at the output of the second circuit And 17 only then when counter 16 counts Q consecutive clock pulses. A discrete sample of the ECS, which coincides with the position on the time axis of the output signal of the And 17 circuit, is taken as the reference count in this cardiocycle, and the signal from the output of the And 17 circuit, which is the output signal of the reference sampling unit 3, is fed to the first input of the memory unit 4, allowing memorization of the reference sample and the following 2p samples of the EX.

Thus, the signal from the output of the block 3 allocation of reference samples controls the process of starting memorization of a group of N = 2n + l samples of ECS, and the first sample in this group, which coincides in time with the mentioned signal, is taken as the reference sample in this cardiocycle.

The number Q is chosen so that it can be achieved by counting consecutive clock pulses only on the TP segment of the electrocardiogram 7. The longest after the TP segment is the ST segment. On the time interval T _ST equal to the duration of the ST segment, q _st = T _st / t fits the time intervals equal to the sampling period T. Thus, if Q> Q _ST +1 is selected, then the counter can count Q consecutive clock pulses only on the TR segment. On all other segments (PQ, ST) of the cardiosignal, the second-order difference samples will go beyond the threshold levels before the counter 16 counts to Q. At the same time, a low level signal will appear at the output of the comparison circuit 13 (Fig. 7a), which sets the counter by the input of R to the zero state and prohibits the passage of clock pulses through the first circuit And 15. When the next input signals of the differences of the second order in the zone between threshold levels, the counter begins to count clock pulses from the beginning. In a similar manner, the device described in the patent of the Russian Federation on the application No. 2002101968/14.

The stored values of N samples from the outputs of the storage unit 4 are sent to the corresponding inputs of the scaling unit 5. The central sample of a group of 2n + 1 samples is transmitted to the output of the scaling unit 5 with a coefficient of 1, and the samples of each of the n pairs of samples located symmetrically relative to the central sample are transmitted with coefficients K _j , j = 1,2, ..., n. At the output of the scaling unit 5, a signal is formed consisting of a sequence of 2n + 1 pulses adjacent to each other, the central pulse has an amplitude equal to the value of the noise at the same moment in time, and for each pair of pulses located to the left and right of the central one, the amplitudes are equal to interference at the appropriate time multiplied by the scale factor K _j . An example of such a signal Uot for n = 3 is shown in Fig. 4, b. In Fig. 4, e, one of the groups of the signal of the Notes is shown on a larger scale. As noted earlier, the spectrum of this signal contains a component at the interference frequency and does not contain the first three spectral zones. In the example, the interference frequency is 0.8 Hz and is 0.8 of the heart rate, taken equal to 60 beats / minute, the frequency of the left side component is (n + 1) -th, i.e. fourth, the spectral zone is 3.2 Hz (see figure 2, c). The signal from the output of the scaling unit 5 is fed to the input of the low-pass filter 6, at the output of which the components of the zero spectral zone are allocated, in our case, a signal is proportional to the drift of the isoline. Since the suppression of the spectral bands is a reduction of energy in the zero spectral components zone, the signal output from lowpass filter 6 is supplied to an amplifier 7. To recover the amplitude detected signal to drift isolines Uiz original signal amplitude drift contour, the gain K of the amplifier 7 _yc must be selected from the condition

where K _f - the transmission coefficient of the low-pass filter at zero frequency.

The signal at the output of amplifier 7 Uiz is shown in FIG. 4c. This signal is fed to one of the inputs of the subtraction unit 8. The signal of the EX-mixture and the drift of the isoline, delayed in the delay unit 9 by a time equal to the delay time of the low-pass filter 6 (EX-signal 1 in Fig. 4d), is received at the second input of the subtraction unit. In the subtraction unit 8, the extracted isoline drift signal is subtracted from the original signal, and an EX signal is generated at the output of this block, cleared of the isoline drift signal (EXo in Fig. 4e).

The technical and economic effect of the proposed method and device for its implementation is that it allows to eliminate the drift of the isoline of the ECS when the frequency of the latter changes from zero to frequencies comparable to and even higher than the heart rate. This helps to improve the conditions for the subsequent processing of the cardiosignal (calculation of the time and amplitude parameters of individual elements of the cardiosignal).

Sources of information

1. US patent No. 6381493 Ischemia detection during nonstandard xardiac excitation patterns.

2. Publication No. 2001121142, A 61 B 5/0444, 20030520.

3. Cardiomonitors. ECG continuous monitoring equipment / A.L. Baranovsky, A.I. Kalinichenko, L.A. Manilo, etc. Ed. A.L. Baranovsky and A.P. Nemirko. -M .: Radio and communications, 1993. S.198 ... 200.

4. Makolkin V.I., Podzolkov V.I., Samoilenko V.V. ECG: analysis and interpretation. -M.: Publ. House “GEOTAR-MED”, 2001. P.27-30.

5. RF patent No. 2195164, A 61 B 5/02. A method for isolating the beginning of a cardiocycle and a device for its implementation, A. A. Mikheev / BI 2002. No. 36.

Claims (2)

1. The way to eliminate the drift of the isoline of the electrocardiogram, which is that the original electrocardiogram (EX) is filtered from high-frequency noise and interference of industrial frequency, sampled, the reference sample is extracted in each cardiocycle, and the drift signal is restored based on the sequence of reference samples using spline approximation contours and subtract this signal from the original signal EX, characterized in that the reference sample is allocated on the TP-segment in each cardiocycle, starting from this reference, remember values of neighboring N = 2n + 1 samples, where n = 1, 2, ..., also lying on the TR segment, the remaining samples within the cardiocycle are equal to zero, the value of the central sample of the group of (2n + 1) samples is left unchanged, and the values of the samples of each of the n pairs of samples located symmetrically with respect to the central sample are scaled in accordance with the condition U _j = U ₀ K _j , where U ₀ is the amplitude of the central sample, U _j is the amplitude of the samples of the j-th sample pair, j = 1, 2, ..., n - number of each pair of samples relative to the central frame, K _j - scale coefficients s, defined as the solution of the system

where τ is the duration of the counting of the EX, T is the sampling period, the obtained sequence of groups of ECS samples is passed through a low-pass filter and then amplified, the obtained contour drift signal is subtracted from the original EX-signal, which is previously delayed by the delay time of the low-pass filter. 2. Device for eliminating the drift of the isoline of the electrocardiosignal, comprising a series-connected low-pass filter, a sampling unit and a block for allocating reference samples, as well as a subtraction unit, characterized in that it also includes a unit for storing values of N samples, a scaling unit, and a second low-pass filter , an amplifier and a delay unit, and the output of the reference sample allocation unit is connected to the first input of the storage unit, the second input of which is connected to the output of the sampling unit, each of N output in the storage unit is connected to one of the N inputs of the scaling unit, the output of the scaling unit is connected to the input of the second low-pass filter, and the input of the latter is connected to the input of the amplifier, the output of which is connected to the first input of the subtraction unit, the second input of the subtraction unit is connected to the output of the delay unit, the input of which is connected to the output of the first low-pass filter, the output of the subtraction block is the output of the device.

Images