KR100400212B1 - Electrocardiogram apparatus having arrhythmia detecting function - Google Patents

Abstract

PURPOSE: An electrocardiogram apparatus having anarrhythmia detecting function is provided to improve the simplicity of a disease diagnosis related to a heart by generating an alarm when an electrocardiogram signal having an abnormal waveform is detected. CONSTITUTION: An analog/digital converter(100) converts an analog electrocardiogram signal into a digital electrocardiogram signal. A digital signal processor(110) eliminates noise from the digital electrocardiogram signal converted by the analog/digital converter(100). A linear prediction filter(120) inputs the digital electrocardiogram signal without the noise to predict a present inputted signal by a previous inputted signal and generates an error signal between the predicted value and the present inputted signal. A signal analyzer(130) detects an arrhythmia signal by the digital electrocardiogram signal without the noise and the error signal generated in the linear prediction filter(120) and generates information required for the diagnosis.

Description

ECG device with arrhythmia detection

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electrocardiogram device which detects an electrocardiogram signal using an electrode on a surface of a body, and amplifies and digitally converts the signal to a computer. The electrocardiogram of an abnormal waveform (arrhythmia, arrythmia) during the progress of a stable electrocardiogram waveform is particularly preferred. The present invention relates to an electrocardiogram device that detects and detects a signal in real time when a signal is detected.

In a general electrocardiogram device, an electrocardiogram signal is sampled at a predetermined period, digitally converted, and input to a signal processing device. The ECG signal input to the signal processing apparatus is corrected through appropriate digital processing, and then recorded in the recording apparatus or used by a diagnostic program to diagnose the condition of the heart. On the other hand, depending on the disease of the heart, in some cases the normal waveform signal progresses, irregular and intermittent abnormal waveforms may occur. The ECG signal generated at this time is an essential signal for diagnosis, and also informs the risk inherent in the heart. . For example, ventricular diastolic (PVC), atrial diastolic (PAC), and crystalline early contractile (PNC) are typical cases, and in order to diagnose such a disease, it is possible to detect a change in minute waveforms that occur during a normal time course. Should be. However, it is difficult for the person who makes the diagnosis to monitor the ECG signal for a long time with attention, and it is more difficult to find the minute change during the long time monitoring. Therefore, when an abnormal waveform occurs, an automatic diagnosis device for automatically recognizing and giving a warning is required.

As described above, a conventional method used for detecting, alarming or diagnosing an abnormal signal appearing intermittently is to store a large amount of data obtained during a time when an abnormal waveform is expected to be processed and processed by a computer or the like. Therefore, the conventional method takes a long time to diagnose, the result is known after the occurrence of the abnormal signal and a predetermined time has elapsed, there is a disadvantage that the diagnostic system is complicated. In addition, the above method cannot be used for a patient monitoring device or the like requiring immediate action.

The present invention was devised to solve the above problems, and when the ECG signal of the abnormal waveform is detected during the progress of the stable ECG waveform, it is detected in real time to generate a predetermined alarm to facilitate diagnosis of diseases related to the heart. It is an object of the present invention to provide an electrocardiogram device having an arrhythmia detection function.

1 is a block diagram showing the configuration of an electrocardiogram device having an arrhythmia detection function according to the present invention.

Figure 2 shows the structure of the linear prediction filter used in the present invention.

Figure 3a is an example 1 of the experimental results in the electrocardiogram device according to the present invention (when the PAC, PNC signal is inserted).

Figure 3b is an example 2 (at the time of insertion of the PVC signal) of the experimental results in the ECG device according to the present invention.

3C is an example 3 (when the heart rate cycle is changed) of the experimental results in the ECG device according to the present invention.

In order to achieve the above object, the electrocardiogram device according to the present invention for extracting the information necessary for diagnosis by analyzing the analog electrocardiogram signal sensed on the body surface comprises an analog / digital conversion unit for converting the analog electrocardiogram signal into a digital electrocardiogram signal; A digital signal processor for removing noise from the digital electrocardiogram signal converted by the analog / digital converter; A linear prediction for inputting a digital electrocardiogram signal from which the noise is removed by the digital signal processor, predicting a signal currently input by a signal input in the past, and generating an error signal between the predicted value and the currently input signal filter; And a signal analyzer configured to detect an arrhythmia signal by the digital electrocardiogram signal from which the noise is generated by the digital signal processor and the error signal generated by the linear prediction filter, and generate information for diagnosis.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing the configuration of an electrocardiogram device having an arrhythmia detection function according to the present invention. 1, an electrocardiogram device having an arrhythmia detection function according to the present invention includes an analog / digital converter 100, a digital signal processor 110, a linear predictive filter 120, and a signal analyzer 130. do.

The analog / digital converter 100 converts an analog electrocardiogram signal into a digital electrocardiogram signal for easy processing by a digital information processor such as a digital computer.

The digital signal processor 110 performs a real time signal processing such as removing noise from the digital electrocardiogram signal converted by the analog / digital converter 100.

The linear predictive filter 120 inputs a digital electrocardiogram signal from which noise is removed by the digital signal processor 110 to predict a signal currently input by a signal input in the past, and the predicted value and the current input. It is a key component of the present invention to generate an error signal from the signal.

The signal analyzer 130 generates information necessary for diagnosis by the digital electrocardiogram signal from which the digital signal processor 110 removes the noise and the error signal generated by the linear predictive filter 120. The signal analysis unit 130 is a kind of digital information processor that stores the input digital signal and the error signal in a memory for a predetermined time to perform a diagnostic program, or outputs a signal waveform to a printer or a user's screen. Do the work of. In addition, in order to process the error signal according to the present invention, the signal analyzer 130 detects an error signal exceeding a predetermined threshold value from the error signal output from the linear predictive filter 120 to generate an abnormal waveform generation signal. An abnormal waveform detection unit 132 for generating a and an alarm generation unit 134 for generating an alarm by the abnormal waveform generation signal generated by the abnormal waveform detection unit 132 may be provided.

In general, the ECG signal is a stationary process because it has a periodic and constant pattern. Therefore, the characteristics of the signal can be estimated and expressed as several parameters.

를 계산하면 입력신호의 주파수 특성을 알 수 있다.A linear prediction filter is used as an estimation method widely used in the area of digital signal processing. Figure 2 shows the structure of the linear prediction filter used in the present invention. Here, the input is an electrocardiogram signal, and the output is an error between the prediction of the next signal to be input and the actual input signal. When the characteristic of the input signal is composed of a finite frequency component and is periodic, an accurate solution can be obtained by the linear predictive filter structure of FIG. 1, in which case the error signal is zero. In this case, the linear prediction filter is a filter (IIR filter) for generating the same signal as the input signal, and the coefficients represent the characteristics of the input signal. That is, when the transfer function of the linear prediction filter is called H (z), the impulse response becomes the waveform of the input signal.

Calculate the frequency characteristic of the input signal.

However, the actual input signal does not satisfy the above condition. Therefore, the error signal is not zero, and the coefficient is adjusted in an optimal state that usually has a minimum mean square error (MMSE). A solution of least squares error exists when the input process is stationary, and a method for solving the solution has been proposed by Wiener et al. However, the method has a complicated calculation process, making it difficult to process in real time. In addition, it is not applied to process a signal which is a partial stationary which is a general characteristic of an actual signal.

As described above, the electrocardiogram signal may also be viewed as a stationary process as a periodic signal, and an optimal solution of the linear predictive filter may be obtained, and the characteristics of the signal may be estimated using this. In addition, it is also possible to find the abnormal waveform to be implemented in the present invention. That is, the error is increased when a signal different from the signal estimated by the continuous signal continuously and its characteristic is increased, and at this time, it can be reported that the singular waveform appears and alert. For this purpose, the adaptive signal processing method is more effective than the complex least-squares error method, which calculates the solution according to the input signal. Thus, in a preferred embodiment of the present invention, by using the method proposed by the widget (widrow) or the like, it is possible to easily process in real time, and also to process the signal of the partial stationary characteristics.

In the present invention, the adaptive signal processing is implemented using a linear prediction method to determine the characteristics of the signal. The adaptive signal processing method shown in the embodiment is a least mean square (LMS) algorithm. I used it. The linear prediction filter of FIG. 2 uses an LMS algorithm. The detailed description of the LMS algorithm is omitted because it is not the scope of the present invention.

The equation for adjusting the coefficient vector (W) is

When the predicted signal, that is, the output of the filter _is called Y _(n)

to be. Where μ is the convergence constant value, X (k) is the input signal at time k, and N is the order of the filter.

The convergence speed and the misadjustment after convergence are determined by μ, and this value and the order (N) of the filter should be used according to the sampling frequency of the ECG signal and the frequency of the abnormal waveform to be detected.

When a high sampling frequency is used as the input of the linear prediction filter, it is difficult to detect an abnormal waveform that appears in the form of low frequency unless the filter order is increased. On the other hand, the input signal of the linear predictive filter can maintain the stationary characteristics even when aliasing occurs when a low sampling rate is used, so that the signal can be estimated, the processing time is free, and the order of the filter Is advantageous because it does not have to be high. However, at this time, since the abnormal waveform in which the high frequency component is weakly changed is not detected, the characteristic of the abnormal waveform to be detected should be considered.

Figure 3b is an example 2 (at the time of insertion of the PVC signal) of the experimental results in the ECG device according to the present invention.

3C is an example 3 (when the heart rate cycle is changed) of the experimental results in the ECG device according to the present invention.

3A, 3B, and 3C are examples of experimental results in an electrocardiogram device according to the present invention. 3A, 3B, and 3C show an input signal at an upper end of each arrhythmia, and a lower end shows a change in error occurring at this time.

Referring to the entire system used here according to the block diagram of Figure 1 as follows. The signals shown at the top of FIGS. 3A, 3B, and 3C are outputs of the analog / digital converter 100 of FIG. 1 and are 2 KHz digital electrocardiogram signals in this example. Here, signals of 12 leads are normally outputted, respectively, but only one of them is used for estimation of the signal. In this example, the lead 1 signal was used. Before using the ECG signal input at 2 KHz as the input of the linear prediction filter 120 of FIG. 1, the sampling period is lowered to 50 Hz and used as the input of the linear prediction filter 120. This is because a low sampling frequency is advantageous for the reasons described above. The order of the linear prediction filter 120 using the LMS algorithm was set to 50th order, and the convergence constant value was set to 0.000000002. The convergence rate is low due to the small value of the convergence constant, but the error is small after the convergence, and when the error is changed by the abnormal waveform and the normal waveform appears again, the error value is recovered again within a short time. The value of the convergence constant should be adjusted to suit the characteristics and circumstances of the system. The signals used were all amplified and digitally converted from the signals of the DYNATECH NEVADA medSim300B, an electrocardiogram simulator. Figure 3a is a result of inserting the waveforms of atrial diastolic (PAC) and nodular early contraction (PNC) during the progress of the normal waveform for the experiment. Figure 3b is an embodiment when the ventricular diastolic (PVC) is inserted, Figure 3c shows the error change when the heart rate changes.

If you look at the change in the error signal, you can see that if the arrhythmia signal is inserted when the error has stabilized to near zero as the normal waveform progresses, the error suddenly increases at that point and then approaches zero as the normal waveform progresses. Can be. In this case, the occurrence of the abnormal waveform may be determined using the error signal, or the stability of the ECG signal may be quantified according to the magnitude or the change shape of the error signal. For example, for patients whose heart rate is not constant, the error value will not converge to zero and will maintain a range of values. The abnormal waveform detection unit 132 of FIG. 1 processes the above process. The linear predictive filter 120 in FIG. 1 does not exist separately, and may be included in the signal analyzer 130 or the digital signal processor 110 having a digital signal processing function.

According to the present invention, it is possible to immediately detect the occurrence of the abnormal waveform with a small amount of calculation, and to measure the stabilization degree of the electrocardiogram by looking at the change state of the error signal or the magnitude of the error value. Specifically, the present invention is to detect a disease in which abnormal waveforms occur intermittently while the normal waveform lasts for a long time, or to detect the change in the period of the electrocardiogram or when the waveform is not constant and wants to know the degree of change is severe. It is used in a small amount of calculation and does not need to examine a large amount of data compared to each other, so that it is easy to process in real time, and when used in a patient monitoring device, there is an effect of preventing the risk by early detection of abnormal waveforms.

However, in the present invention, when the abnormal waveform is periodically maintained, since the linear predictive filter has an optimal value in a state suitable for the characteristics of the signal, the error value is close to 0, and thus there is a problem in that information on the disease cannot be obtained. .

Claims (3)

In the electrocardiogram device for extracting the information necessary for diagnosis by analyzing the analog electrocardiogram signal detected on the body surface, An analog / digital converter converting the analog ECG signal into a digital ECG signal; A digital signal processor for removing noise from the digital electrocardiogram signal converted by the analog / digital converter; A linear prediction for inputting a digital electrocardiogram signal from which the noise is removed by the digital signal processor, predicting a signal currently input by a signal input in the past, and generating an error signal between the predicted value and the currently input signal filter; And An arrhythmia detection function including a digital electrocardiogram signal from which noise is removed by the digital signal processor and a signal analyzer for detecting arrhythmia signal by the error signal generated by the linear prediction filter and generating information for diagnosis ECG device with a. The method of claim 1, wherein the signal analysis unit An abnormal waveform detection unit for generating an abnormal waveform generation signal by detecting an error signal exceeding a predetermined threshold value from the error signal output from the linear prediction filter; And EKG having an arrhythmia detection function characterized in that it comprises an alarm generation unit for generating an alarm by the abnormal waveform generation signal generated by the abnormal waveform detection unit. The linear predictive filter of claim 1, wherein the linear predictive filter comprises: An electrocardiogram device having an arrhythmia detection function using a least mean square algorithm as a method of predicting a signal currently input by a signal input in the past.

Images