KR102451623B1 - Method and Apparatus for Comparing Features of ECG Signal with Difference Sampling Frequency and Filter Methods for Real-Time Measurement - Google Patents

Abstract

Disclosed are a method and an apparatus for comparing the features of an electrocardiogram (ECG) signal by using different sampling frequencies and filter methods for real-time measurement. The method proposed in the present invention comprises the steps of: setting the parameters of a high-pass filter and a low-pass filter including the order and movement delay interval of a filter to remove the noise of an ECG signal; extracting an ECG signal by using a plurality of different sampling frequencies; and performing comparison and analysis on the extracted ECG signal with respect to the plurality of different sampling frequencies before and after the high-pass filter and the low-pass filter are applied to the extracted ECG signal. Therefore, the method can analyze important components of the ECG signal such as noise factors that affect a method for designing a signal and noise cancellation filter.

Description

Method and Apparatus for Comparing Features of ECG Signal with Difference Sampling Frequency and Filter Methods for Real-Time Measurement

The present invention relates to a method and apparatus for comparing characteristics of an ECG signal using a difference sampling frequency and a filter technique for real-time measurement.

According to WHO statistics, the number of people with cardiovascular disease, especially the elderly, is increasing [1]. Therefore, monitoring of cardiovascular problems is becoming more important [2-4]. Electrocardiography is a non-invasive measurement method commonly used for screening and diagnosing various diseases [5-8]. It is one of the important applications in condition monitoring and diagnosis [9, 10], and is one of the essential information to support stroke diagnosis. Such ECG monitoring helps to detect and identify the cause of stroke at an early stage [11]. The quality of the ECG signal plays a direct role in the diagnosis [12]. Components of the ECG signal include Q, R, S, and T parts [13]. Here, the R peak plays an essential role in calculating the patient's heart rate [9]. In this way, more useful information can be obtained by classifying the components of the ECG signal [14].

In general, ECG recording is made by attaching electrodes to the patient's body, and the device receives electrical signals from these electrodes [15]. Therefore, the contact between the electrode and the user's skin directly affects the quality of the acquired ECG signal. Also, when a user is near equipment that uses alternating current (AC) power, the human body is affected by interference from the power grid [16-18]. This interference has the same frequency as the 50/60 Hz grid frequency depending on the region [19]. These are two types of noise that directly affect the quality of the received ECG signal. Therefore, it is necessary to eliminate these two types of interference [20-22]. There are two methods of noise removal: an analog filter designed in hardware and a digital filter designed in software. The analog filter consists of three basic elements, a resistor (R), an inductor (L), and a capacitor (C), and has fixed parameters [15, 23]. Since digital filters are designed based on the computing power of the central processing unit (CPU), parameters can be changed more flexibly [24]. In order to completely remove both types of noise from the ECG signal, both low-pass (LP) and high-pass (HP) filters need to be designed [25]. Therefore, the choice of filter design method plays an essential role in signal quality [16, 22]. Recently, dedicated filter architectures such as adaptive filters using autoencoders or machine learning-based filters have been proposed [26-28].

Along with advanced semiconductor technology, there have been many electromagnetic components from major manufacturers that provide biological induction, such as TI, Maxim Integrated, and Analog Devices, and the smaller and more precise form of these components makes the design of wearable devices faster and easier. [3, 24, 29, 30]. In addition, the streamlined design process accelerates the commercial product launch process [31, 32].

These biosensors generally have variable sampling frequencies [33]. For ECG signals, the higher the sampling frequency, the more accurate the signal and vice versa. However, the high sampling frequency can result in extensive databases that are difficult to compute data. The sampling frequency of the BIT/MIH database is 360 Hz [34]. The sampling frequency of the sensor ADS1293 from TI is up to 2133 Hz [35], the frequency of the sensor MAX86150 from Maxim Integrated is 3200 Hz [36], and the sampling frequency of Analog Devices (ADAS1000) is 8000 Hz. Therefore, irradiation of ECG signals at different sampling frequencies plays an important role in ensuring signal accuracy while reducing the amount of data to be collected. The basic structure of a wearable device is a biosensor, a microcontroller unit (MCU), Bluetooth, Wi-Fi, and a battery management unit [33, 37-39]. Since most ECG wearable devices use batteries, reducing the amount of data computing is also an effective way to save energy consumption and extend usage time [24, 33].

An object of the present invention is to provide a method and apparatus for analyzing important components of an electrocardiogram signal, such as a noise factor affecting a signal and a method of designing a noise removal filter. According to an embodiment of the present invention, an electrocardiogram signal of a BIT/MIH database is analyzed at different sampling frequencies, and a filter is applied to compare the difference between the received signals. According to an embodiment of the present invention, an ECG signal acquisition model based on mass production factors is proposed to collect real-time signals, and filters are applied to remove noise with different sampling frequencies of the device, and parameters are compared. I want to analyze

In one aspect, the method for comparing the characteristics of an ECG signal using different sampling frequencies and filter techniques for real-time measurement proposed by the present invention is a high-pass including a filter order and a movement delay interval for removing noise from the ECG signal. Before and after the steps of setting parameters of a filter and a low-pass filter, extracting an ECG signal using a plurality of different sampling frequencies, and applying the high-pass filter and the low-pass filter to the extracted ECG signal, and the and comparing and analyzing a plurality of different sampling frequencies.

The step of setting the parameters of the high-pass filter and the low-pass filter including the order of the filter for removing the noise of the ECG signal and the movement delay interval is an Infinite Impulse Response (IIR) filter and a symmetrical finite impulse response ( Finite Impulse Response (FIR) filter parameters are set respectively, and the symmetric finite impulse response filter has an integer delay when the coefficient is odd, and the order of the infinite impulse response filter is smaller than the coefficient of the symmetric finite impulse response filter. set, and only the infinite impulse response filter is applied to the high-pass filter.

In the step of extracting the ECG signal using the plurality of different sampling frequencies, data about the ECG signal extracted with the sampling frequency in the BIT/MIH database is generated as original data, and extracted using the plurality of different sampling frequencies. Data on the acquired ECG signal is generated as comparison data in the MIT/BIT database.

Before and after applying the high-pass filter and the low-pass filter to the extracted electrocardiogram signal, and the step of comparatively analyzing the plurality of different sampling frequencies before and after applying the high-pass filter and the low-pass filter, and the A criterion for comparing the original data and the comparison data includes a signal-to-noise ratio, and compares the amplitudes of the P-wave, QRS-wave, and T-wave of the ECG signal with respect to a plurality of different sampling frequencies.

In another aspect, the apparatus for comparing the characteristics of the ECG signal using different sampling frequencies and filter techniques for real-time measurement proposed by the present invention includes the order and movement delay interval of the filter for removing the noise of the ECG signal. A filtering unit including a high-pass filter and a low-pass filter applied by setting parameters, a sampling unit for extracting an ECG signal using a plurality of different sampling frequencies, and the high-pass filter and low-pass filter for the extracted ECG signal and an analysis unit that compares and analyzes the plurality of different sampling frequencies before and after applying the filter.

According to embodiments of the present invention, it is possible to analyze important components of an electrocardiogram signal, such as a noise factor affecting a signal and a method of designing a noise removal filter. By applying the filter according to an embodiment of the present invention, it is possible to analyze the electrocardiogram signal of the BIT/MIH database at different sampling frequencies, and compare the difference between the received signals. Through the ECG signal acquisition model based on mass production factors for collecting real-time signals according to an embodiment of the present invention, filters for removing noise at different sampling frequencies may be applied, and parameters may be compared and analyzed.

1 is a view for explaining an electrocardiogram signal component according to the prior art.
2 is a flowchart illustrating a method for comparing characteristics of an ECG signal using different sampling frequencies and filter techniques for real-time measurement according to an embodiment of the present invention.
3 is a diagram illustrating a configuration of an apparatus for comparing characteristics of an ECG signal using different sampling frequencies and filter techniques for real-time measurement according to an embodiment of the present invention.
4 is a diagram for explaining the configuration of a filtering unit according to an embodiment of the present invention.
5 is a graph comparing ECG signals having different low sampling frequencies according to an embodiment of the present invention.
6 is a graph comparing ECG signals having different sampling frequencies for P, QRS, and T wave detection according to an embodiment of the present invention.
7 is a graph illustrating an ECG signal and a filtered ECG signal in a time domain according to an embodiment of the present invention.
8 is a diagram showing the configuration of a real-time ECG signal measurement system according to an embodiment of the present invention.
9 is a graph illustrating an electrocardiogram signal having different sampling frequencies before and after filtering in the time domain according to an embodiment of the present invention.
10 is a graph illustrating an electrocardiogram signal having different sampling frequencies before and after filtering in the time domain according to another embodiment of the present invention.

Electrocardiogram (ECG) signals have been used to monitor and diagnose signs of cardiovascular disease and abnormal signals in the human body. The ECG signal is typically characterized by PR, QRS, QT interval, ST-segment and Heart Rate (HR) parameters. Electrocardiogram devices are widely used in many applications, especially in the elderly. However, the ECG signal is often affected by environmental noise. There are two types of noise that affect the ECG signal: low frequencies, which originate from muscle activity, and 50/60 Hz, which originates from the electrical grid. Removal of these noises is important for improving the quality of the ECG signal. If the ECG signal is clear by removing the noise, cardiovascular disease can be easily diagnosed. ECG signals with a higher sampling frequency are more accurate. However, it is clearer and more difficult to remove this noise with a filter on the signal. Therefore, it is necessary to analyze the symmetrical correlation between the signal sampling frequency, signal to noise ratio (SNR), and signal parameters such as signal amplitude.

In the present invention, undesired ECG noise is filtered by using a simple filter phase to reduce design complexity. In the present invention, characterization evaluation of electrocardiogram signals performed at different sampling frequencies before and after applying an infinite impulse response (IIR) filter and a symmetric finite impulse response (FIR) filter is compared. Therefore, for accurate diagnosis, the sampling frequency must be consistent with the same frequency of the ECG signal. In addition, approaches that help reduce the computing power and hardware resources of the device can also be important.

Results according to an embodiment of the present invention were tested using the MIT/BIH database at a sampling frequency of 360 Hz with 11-bit resolution. In addition, a device operating in real time with a sampling frequency of 100 Hz to 2133 Hz with 24-bit resolution was tested. The test results according to the embodiment of the present invention show the advantages of the symmetric FIR filter compared to the IIR when applied to the electrocardiogram signal filtering. Therefore, the quality of the ECG signal can be improved by applying the proposed method and apparatus. Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

In general, there is a correlation between sampling frequency and signal quality. The higher the sampling frequency, the more reliable the components of the signal. However, ECG signal parameters should be considered as heart rate (eg 60-150 beats/min) and P, QRS, and T wave amplitudes. In the present invention, a conclusion was drawn by analyzing the asymmetry of the ECG signal at a plurality of different frequencies. The noise factor affects the signal and how the denoising filter is designed. In the present invention, important components of such an electrocardiogram signal are analyzed.

According to an embodiment of the present invention, an electrocardiogram signal of a BIT/MIH database is analyzed at different sampling frequencies, and a filter is applied to compare the difference between the received signals.

According to an embodiment of the present invention, an ECG signal acquisition model based on mass production factors is proposed to collect real-time signals. In addition, filters are applied to remove noise at different sampling frequencies of the device, and parameters are compared and analyzed.

1 is a view for explaining an electrocardiogram signal component according to the prior art.

Figure 1 shows a typical electrocardiogram signal including a P wave, a QRS complex wave, and a T wave [40]. P waves are caused by muscle activation in both hearts. Left-right ventricular activation produces a QRS complex wave. Finally, activation of the ventricular muscle generates T waves. Therefore, P, QRS, and T waves are essential elements to be observed and evaluated because they convey a lot of useful information [13, 41].

According to the prior art, an arrhythmia can be identified by analyzing the shape characteristics of the electrocardiogram P-QRS-T segment and P-QRS-T wave [2]. Another prior art used a novel neural network structure based on the 12 most common electrocardiogram leads proposed to classify 9 arrhythmias [5]. Another prior art used a time-independent QRS detection method [10]. Graph data representation was applied, and a heterogeneous time scale transformation was applied to find the exact P, QRS, and T wave breakpoints. Another prior art compared modeling and segmentation of P and T waveforms in electrocardiograms using mathematical models of Gaussian and Rayleigh functions [13]. Root mean square (RMS) evaluated the fit between each model and each feature waveform. In addition, in the prior art, wave transformation was applied to estimate P-wave and T-wave. In another prior art, a study on the structure of an ECG signal acquisition system including hardware and software was initiated [24]. Here, factors such as noise reduction, machine learning, and privacy protection related to software are analyzed. As a result, detailed information about the components of the ECG signal was provided.

Among these prior art techniques, there is no technique focused on analyzing the sample frequency and signal quality of the ECG signal. Therefore, the present invention proposes a new method related to ECG signal acquisition and analysis.

2 is a flowchart illustrating a method for comparing characteristics of an ECG signal using different sampling frequencies and filter techniques for real-time measurement according to an embodiment of the present invention.

The proposed method for comparing the characteristics of the ECG signal using different sampling frequencies and filter techniques for real-time measurement is to measure the parameters of the high-pass filter and low-pass filter, including the order and movement delay interval of the filter to remove the noise of the ECG signal. Before and after the setting step 210, extracting an electrocardiogram signal using a plurality of different sampling frequencies 220, and applying the high-pass filter and the low-pass filter to the extracted electrocardiogram signal, and the plurality of and comparing and analyzing (230) different sampling frequencies.

In step 210, parameters of the high-pass filter and the low-pass filter including the order of the filter for removing noise from the ECG signal and the movement delay interval are set.

The high-pass filter and the low-pass filter according to an embodiment of the present invention include an Infinite Impulse Response (IIR) filter or a symmetric Finite Impulse Response (FIR) filter. In step 210, parameters for the IIR filter and the symmetric FIR filter are respectively set. A characteristic of the symmetric FIR filter is that it is always stable and the signal is not distorted. However, symmetric FIR has a significant amount of computation, so it must be appropriately selected for its intended use. Also, a symmetric FIR filter with odd coefficients has an integer delay. The order of the IIR filter is set to be smaller than the coefficient of the symmetric FIR filter, and since the signal passing through the IIR filter is distorted, it should be designed to ensure stability.

Only the infinite impulse response filter is applied to the high-pass filter according to the embodiment of the present invention. FIR filters require dedicated calculations. Therefore, it is not appropriate to apply the FIR filter to the high-pass filter. Therefore, in the present invention, only the IIR filter is used for the high-pass filter.

The purpose of the low-pass filter is to remove various harmonics and ground-bouncing noise from the power line. The noise frequency is equal to the frequency of the power grid, which is 50/60 Hz, depending on the system's power supply network. The low-pass filter according to an embodiment of the present invention includes a combination of an IIR filter and a symmetric FIR filter.

In step 220, an electrocardiogram signal is extracted using a plurality of different sampling frequencies.

First, the data on the ECG signal extracted with the sampling frequency in the BIT/MIH database is generated as original data, and the data on the ECG signal extracted using a plurality of different sampling frequencies are compared in the MIT/BIT database as data. create The sampling frequency of the BIT/MIH database data is 360 Hz, and a plurality of different sampling frequencies according to an embodiment of the present invention may include 30, 45, 72, 90, 120, and 180 Hz. Such a plurality of different sampling frequencies is only an example, but is not limited thereto, and various frequencies may be applied.

In step 230, before and after applying the high-pass filter and the low-pass filter to the extracted electrocardiogram signal, and the plurality of different sampling frequencies are compared and analyzed.

Data extracted with a sampling frequency of 360 Hz is the original data, and a plurality of different sampling frequencies according to an embodiment of the present invention are selected by filtering data extracted at 30, 45, 72, 90, 120, 180 Hz as comparison data. do. Criteria for comparing the original data with the comparison data before and after applying the high-pass filter and the low-pass filter include a signal-to-noise ratio. Thereafter, the amplitudes of the P wave, QRS wave, and T wave of the ECG signal with respect to a plurality of different sampling frequencies are compared.

3 is a diagram illustrating a configuration of an apparatus for comparing characteristics of an ECG signal using different sampling frequencies and filter techniques for real-time measurement according to an embodiment of the present invention.

The proposed apparatus for comparing characteristics of an electrocardiogram signal using different sampling frequencies and filter techniques for real-time measurement includes a filtering unit 310 , a sampling unit 320 , and an analysis unit 330 .

The filtering unit 310 according to an embodiment of the present invention sets parameters of the high-pass filter and the low-pass filter including the order and movement delay interval of the filter for removing noise from the ECG signal.

The high-pass filter and the low-pass filter of the filtering unit 310 according to an embodiment of the present invention include an infinite impulse response (IIR) filter or a symmetric finite impulse response (FIR) filter. The filtering unit 310 sets parameters related to the IIR filter and the symmetric FIR filter, respectively. A characteristic of the symmetric FIR filter is that it is always stable and the signal is not distorted. However, symmetric FIR has a significant amount of computation, so it must be appropriately selected for its intended use. Also, a symmetric FIR filter with odd coefficients has an integer delay. The order of the IIR filter is set to be smaller than the coefficient of the symmetric FIR filter, and since the signal passing through the IIR filter is distorted, it should be designed to ensure stability.

Only the infinite impulse response filter is applied to the high-pass filter according to the embodiment of the present invention. FIR filters require dedicated calculations. Therefore, it is not appropriate to apply the FIR filter to the high-pass filter. Therefore, in the present invention, only the IIR filter is used for the high-pass filter.

The purpose of the low-pass filter is to remove various harmonics and ground-bouncing noise from the power line. The noise frequency is equal to the frequency of the power grid, which is 50/60 Hz, depending on the system's power supply network. The low-pass filter according to an embodiment of the present invention includes a combination of an IIR filter and a symmetric FIR filter. The filtering unit 310 according to an embodiment of the present invention will be described in more detail with reference to FIG. 4 .

4 is a diagram for explaining the configuration of a filtering unit according to an embodiment of the present invention.

A high-pass filter and a low-pass filter should be designed to remove high-pass noise and low-pass noise from the grid for ECG signals [42]. FIR and IIR filters are commonly used for two types of filtering. Equations (1) and (2) describe the FIR filter and the IIR filter, respectively.

(1)

(One)

(2)

(2)

Here, y(n) is an output signal (that is, an electrocardiogram signal after filtering) 431, x(n) is an input signal (that is, an electrocardiogram signal before filtering) 432, and n is a filter order.

A characteristic of a symmetric FIR filter is that it is always stable and the signal is not distorted. However, symmetric FIR has a significant amount of computation, so it must be appropriately selected for its intended use. Also, a symmetric FIR filter with odd coefficients has an integer delay. In other words, it means the original signal and the filtered signal simply by moving the integer interval. This shift interval can be expressed by the following formula:

(3)

(3)

Here, n is a filter order and d is a movement delay interval.

The order of the IIR filter according to the embodiment of the present invention is much smaller than that of the FIR filter. However, since the signal passing through the IIR filter is distorted, it must be designed to ensure stability.

According to an embodiment of the present invention, noise is removed with a cutoff frequency of 0.01 Hz to 0.05 Hz by applying a high-pass filter to the ECG signal. FIR filters require dedicated calculations. Therefore, it is not appropriate to apply the FIR filter to the high-pass filter. Only the IIR filter is used for the high pass filter 420 according to the embodiment of the present invention.

The purpose of the low-pass filter is to remove various harmonics and ground-bouncing noise from the power line. The noise frequency is equal to the frequency of the power grid, which is 50/60 Hz, depending on the system's power supply network. According to an embodiment of the present invention, a filter structure is proposed as shown in FIG. 4 . The structure of the filtering unit according to the embodiment of the present invention includes a combination of a low-pass filter (IIR 411 and FIR 412) and a high-pass filter (IIR 420).

According to an embodiment of the present invention, an IIR filter is designed according to the Elliptic method in order to minimize the number of filter calculations, and an FIR filter is designed according to the Equiripple method.

Referring back to FIG. 3 , the sampling unit 320 according to an embodiment of the present invention extracts an electrocardiogram signal using a plurality of different sampling frequencies.

First, the data on the ECG signal extracted with the sampling frequency in the BIT/MIH database is generated as original data, and the data on the ECG signal extracted using a plurality of different sampling frequencies are compared in the MIT/BIT database as data. create The sampling frequency of the BIT/MIH database data is 360 Hz, and a plurality of different sampling frequencies according to an embodiment of the present invention may include 30, 45, 72, 90, 120, and 180 Hz. Such a plurality of different sampling frequencies is only an example, but is not limited thereto, and various frequencies may be applied.

The analysis unit 330 according to an embodiment of the present invention compares and analyzes the plurality of different sampling frequencies before and after applying the high-pass filter and the low-pass filter to the extracted ECG signal.

Data extracted with a sampling frequency of 360 Hz is the original data, and a plurality of different sampling frequencies according to an embodiment of the present invention are selected by filtering data extracted at 30, 45, 72, 90, 120, 180 Hz as comparison data. do. Criteria for comparing the original data with the comparison data before and after applying the high-pass filter and the low-pass filter include a signal-to-noise ratio. Thereafter, the amplitudes of the P wave, QRS wave, and T wave of the ECG signal with respect to a plurality of different sampling frequencies are compared.

As described above, the sampling frequency of the BIT/MIH database data is 360 Hz. In an embodiment of the present invention, the ECG signal extracted from this database is analyzed with a smaller sampling frequency. A plurality of different sampling frequencies according to an embodiment of the present invention may be selected as follows: f _{new_fre} = (30, 45, 72, 90, 120, 180) Hz.

The evaluation of data generated in the BIT/MIH database is as follows. First, a new database is created from the MIT/BIT database using the f _{new_fre} sampling frequency. Filters are performed on both new and original data.

Data extracted from a sampling frequency of 360 Hz is selected as original data, and data extracted from a plurality of different sampling frequencies is filtered and selected as comparison data.

The comparison criterion includes a signal to noise ratio (SNR) of a signal, and compares the amplitudes of the P wave, QRS wave, and T wave at a plurality of different sampling frequencies.

The SNR of the signal before and after applying the filter can be expressed as:

(4)

(4)

According to Equations (1) and (2), the total number of multiplication and addition calculations of the FIR/IIR filter is as follows.

(5)

(5)

(6)

(6)

The total number of multiplications and operations performed by the FIR/IIR filter on the entire data is:

(7)

(7)

(8)

(8)

Here, L is the data length, M is the order of the FIR, and N is the order of the IIR filter.

Next, the filtering effect on the ECG signal at a plurality of different sampling frequencies is evaluated. According to an embodiment of the present invention, ECG signals from the MIT/BIH database were used, and the signals were collected directly from voluntary participants. ECG samples from the MIT/BIH database are fixed at a sampling rate of 360 Hz. However, the ECG signals of voluntary participants had a dynamic sampling frequency of 100 Hz to 2133 Hz.

5 is a graph comparing ECG signals having different low sampling frequencies according to an embodiment of the present invention.

5(a) is original data, and FIGS. 5(b) to 5(d) are graphs illustrating ECG signals having sampling frequencies of 30, 45 and 60 Hz, respectively.

According to an embodiment of the present invention, a high-pass filter and a low-pass filter were evaluated using five electrocardiogram signals from the MIT/BIH database. Then, using these data, seven new ECG signals are generated with sampling frequencies of 30, 45, 60, 72, 90, 120, and 180 Hz. For the proposed design, in the present invention, the lowest possible sampling rate suitable for heart rate monitor applications is selected because reducing the size of the data can increase the system's computational efficiency and overall power overhead. The minimum sampling frequency is suitable for ECG-related applications where P-wave, QRS, and P-wave parameters need to be evaluated.

In addition, five ECG signals from the MIT/BIH database with sample frequencies of 30, 45, and 60 Hz were used. The results show that for the 30 Hz frequency, the data for the R peak of the signal is lost significantly, so it is not suitable for applications to calculate the heart rate using the R peak data of the ECG signal. For a sampling frequency of 45 Hz, there is still a loss of peak R. Therefore, the minimum sampling frequency of 45 Hz is suitable for applications using the R peak of the ECG signal.

6 is a graph comparing ECG signals having different sampling frequencies for P, QRS, and T wave detection according to an embodiment of the present invention.

6(a) is original data, and FIGS. 6(b) to 6(d) are graphs illustrating ECG signals having sampling frequencies of 90, 120 and 180 Hz, respectively.

According to an embodiment of the present invention, five ECG signals having sample frequencies of 60, 72, 90, 120, and 180 Hz are examined in the MIT/BIH database. The results show that the determinant Q and S values can generally be wrong for a frequency of 90 Hz. Therefore, we choose the minimum sampling frequency value suitable for applications that consider P, QRS, and T waves to be 90 Hz.

7 is a graph illustrating an ECG signal and a filtered ECG signal in a time domain according to an embodiment of the present invention.

7 shows the original ECG signal in the time domain and noise spectrum with a sampling rate of 360 Hz. Fig. 7(a) shows the electrocardiogram signal in the time domain, and Fig. 7(b) shows that filtered signals such as low-pass FIR filter (blue) and low-pass IIR filter (red) provide sufficient noise suppression for the original signal. It clearly shows that the accurate processing of ECG signals can be improved.

Next, the effect of the filter on the ECG signal having sampling frequencies of 90, 120, 180, and 360 Hz is analyzed. In the case of signals with sampling frequencies of 90 and 120 Hz, the low-pass filter is not applied because the noise of the ECG signal is the noise of the power grid with a frequency of 60 Hz. The 90 Hz and 120 Hz sampling frequencies are not sufficient to evaluate the effect of this noise.

Table 1 shows the signal SNR before and after the filter application, the amplitudes of the P, QRS, and T waveforms before and after the filter, and the number of calculations corresponding to each filter.

<Table 1>

It can be seen that changing the sampling frequency changes the SNR of the signal. The SNR of the electrocardiogram signal tends to increase. Therefore, it can be clearly seen that the smaller sampling frequency can reduce the effect of spectral noise of the power delivery network at 60 Hz.

For 90Hz and 120Hz sampling frequencies for ECG signal reconstruction, we only apply the lowpass IIR filter because the lowpass IIR filter provides a lower filtering order and can improve the computational efficiency much higher for the lowpass FIR filter. Because the IR filter distorts the ECG signal and the SNR of the reconstructed ECG signal decreases in the frequency range of 35~45Hz, the SNR of the ECG signal at sampling frequencies of 90Hz and 120Hz is smaller than the SNR of the original signal. In addition, the SNR characteristics according to a wide range of filter combinations at 180Hz and 360Hz sampling frequencies were investigated and analyzed. The SNR of the ECG signal with high-pass filter and low-pass filter IIR/FIR filter combined is high-pass (HP) and low-pass (LP) IIR/FIR, HP-LP (IIR-IIR), HP-LP (IIR-FIR) ), etc., can provide the best performance in all filter combinations. At low sampling frequencies below 120 Hz, there is no need to apply filters to improve the ECG signal.

In terms of amplitude at different ECG sampling frequencies, the ECG signal is 180 Hz and 360 Hz, and the amplitudes of the P, QRS, and T waves of the symmetric FIR filter signal are larger than that of the IIR filter. IIR filters can reduce signal amplitude more than FIR filters, increasing the difficulty of locating peak ECG signals. The FIR filter makes it easy to determine the position and amplitude of the QRS peak using the denoised signal. However, the IIR filter may cause an unwanted ringing signal as shown in FIG. 7(b). Therefore, filter combination and filter order optimization selection are the most important factors in terms of SNR, amplitude and computational complexity.

The results in Table 1 also show the large difference in the number of calculations that must be performed between the IIR filter and the symmetric FIR filter at different frequencies.

At a sampling frequency of 180 Hz, the odd-order symmetric FIR filter has an order of 45 while the IIR filter has an order of 15. However, at different sampling frequencies (i.e. 360 Hz), the FIR filter order must be changed from 45 to 91, which can greatly increase the computational complexity. Therefore, it can be confirmed that the selection of the sampling frequency at 180 Hz may be best suited for the ECG signal, since the low-pass FIR filter can significantly remove the noise of the power delivery network operating at 50 and 60 Hz frequencies.

8 is a diagram showing the configuration of a real-time ECG signal measurement system according to an embodiment of the present invention.

As shown in FIG. 8, a test model was created for measuring the ECG signals of voluntary participants. The hardware architecture includes an electrocardiogram integrated circuit (ADS1293) and an evaluation board (ESP32). The ADS1293 board is connected to the participant via electrodes [43] and communicates with the ESP32 board via the Serial Peripheral Interface (SPI). The connection signals of ESP32 board and ADS1293 module include major components such as (GPIO23-MOSI), (GPIO19-MISO), (GPIO18-SK), (GPIO17-ADS1293_CS), (GPIO16-ADS1293_INT). For the ADS1293 chip, the configured signals include (IN0-RA), (IN1-LA), (IN2-LL), (IN3-RL): The operating configuration and other parameters are described in detail here [19]. The implementation code for this system uses the Arduino library. The sampling frequency of the ADS1293 can be adjusted to suit the test. In the evaluation test, sampling frequencies of the ECG signals were selected as 100 Hz, 200 Hz, 400 Hz, and 2133 Hz, respectively. Different sampling frequencies can be used to monitor and eliminate the effects of different types of noise on the ECG signal. As shown in Fig. 4, the proposed filter architecture was applied to the system to remove unwanted noise from the ECG signals collected from participants. The software for collecting and displaying the ECG signal is done in MATLAB. The key results including SNR, ECG amplitude and utilized filter order are summarized in Table 3.

Table 2 shows the SNR of the ECG signal measured at each sampling frequency of 2133, 400, 200, and 100 Hz.

<Table 2>

(30, 40, 60, 90, 120, 180) Hz can be reconstructed from the original MIB data (330 Hz). A plurality of different sampling frequencies according to an embodiment of the present invention were used for measurement similarly to the reconstructed signal frequency (ie, a 100 Hz signal was measured for the reconstructed 90 Hz). This comes from the fixed frequency limit of the chip being utilized (ie 2333 Hz on the ADS1293).

9 is a graph illustrating an electrocardiogram signal having different sampling frequencies before and after filtering in the time domain according to an embodiment of the present invention.

10 is a graph illustrating an electrocardiogram signal having different sampling frequencies before and after filtering in the time domain according to another embodiment of the present invention.

9(a) and 9(b) respectively show the ECG signal and the spectrum of the ECG signal in the time domain at the 2133 Hz sampling frequency, and FIGS. 9(c) and 9(d) are respectively at the 400 Hz sampling frequency. The ECG signal and the spectrum of the ECG signal in the time domain are shown.

10(a) and 10(b) show the ECG signal and the spectrum of the ECG signal in the time domain at 200 Hz sampling frequency, respectively, and FIGS. 10(c) and 10(d) are each at 100 Hz sampling frequency. The ECG signal and the spectrum of the ECG signal in the time domain are shown.

As a result of analyzing the measured signals such as SNR and amplitude, it was found that no filters were required for P, QRS, and T waves at a sampling frequency of 100 Hz. However, as the sampling frequency increases, it can be seen that the SNR of the original ECG signal is lowered as shown in FIGS. 9(a, c) and 10(a). We found that the amplitude of the symmetric FIR filter was higher than that of the IIR filter in terms of the amplitudes of the P, QRS, and T waves measured in the ECG signal. In terms of the number of filters required, a lowpass IIR filter can provide the same number of filter orders up to 2133Hz. However, when the sampling frequency was increased above 2133 Hz, the required filter order increased significantly.

The device described above may be implemented as a hardware component, a software component, and/or a combination of the hardware component and the software component. For example, devices and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA), It may be implemented using one or more general purpose or special purpose computers, such as a programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, although one processing device is sometimes described as being used, one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that may include For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as parallel processors.

The software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or device, to be interpreted by or to provide instructions or data to the processing device. may be embodied in The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like.

As described above, although the embodiments have been described with reference to the limited embodiments and drawings, various modifications and variations are possible by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

<References>

1. Organization, W. H., Cardiovascular diseases (CVDs). 17 May 2017.

2. Wenliang Zhu, X.C., Yan Wang, Lirong Wang, Arrhythmia Recognition and Classification Using ECG Morphology and Segment Feature Analysis. IEEE/ACM Trans Comput Biol Bioinform, 2019. 16(1): p. 131-138.

3. Magsi, H.S., A.H.; Al-Rakhami, M.S.; Zahid, N.; Pirbhullal, S.; Wang, L., A Novel Adaptive Battery-Aware Algorithm for Data Transmission in IoT-Based Healthcare Applications. Electronics 2021. 10.

4. George, U.Z.M., K.S.; Lee, S. Q., Extraction and Analysis of Respiratory Motion Using a Comprehensive Wearable Health Monitoring System. Sensors 2021. 21.

5. Li, D. W., H.; Zhao, J.; Tao, Y.; Fu, J., Automatic Classification System of Arrhythmias Using 12-Lead ECGs with a Deep Neural Network Based on an Attention Mechanism. Symmetry 2020. 12(11).

6. Li, D. T., Y.; Zhao, J.; Wu, H., Classification of Congestive Heart Failure from ECG Segments with a Multi-Scale Residual Network. Symmetry 2020. 12.

7. Peters, S., Emery-Dreifuss FHL1 Mutation: A Credo for a Better ECG Definition in Arrhythmogenic Cardiomyopathy. Modern Journal of Medicine and Biology, 2021.

8. Bao, X.A., A.K.; Kamavuako, E.N., Estimation of the Respiratory Rate from Localized ECG at Different Auscultation Sites. Sensors 2021. 21.

9. Shanshan Chen, W.H., Zhi Li, JianLi, XingjiaoGao, Heartbeat classification using projected and dynamic features of ECG signal. Biomedical Signal Processing and Control, January 2017. 31: p. 165-173.

10. Augustyniak, P., Diagnostic Interpretation of Non-Uniformly Sampled Electrocardiogram. Sensors 2021. 21.

11. Chen, Y.L.Z., Intelligent and Real-Time Data Acquisition for Medical Monitoring in Smart Campus. IEEE Access 23 November 2018. 6: p. 74836 - 74846.

12. Syed Khairul Bashar, E.D., Allan J. Walkey, David, D. Mcmanus, And Ki H. Chon, Noise Detection in Electrocardiogram Signals for Intensive Care Unit Patients. IEEE Access 01 July 2019. 7: p. 88357 - 88368.

13. Schlindwein, J.P.V.M.E.M.B.S.P.C.C.J.H.S.F.F.S., Evaluating Gaussian and Rayleigh-Based Mathematical Models for T and P-waves in ECG. IEEE Latin America Transactions, 2017. 15(5).

14. J.Mateo, A.M.T., A.Aparicio, J.L.Santos, An efficient method for ECG beat classification and correction of ectopic beats. Computers & Electrical Engineering, July 2016. 53: p. 219-229.

15. Ngoc Thang Bui; Duc Tri Phan; Thanh Phuoc Nguyen; Giang Hoang; Jaeyeop Choi; Quoc Cuong Bui, J.O., Real-Time Filtering and ECG Signal Processing Based on Dual-Core Digital Signal Controller System. IEEE Sensor, 2020. 20(12): p. 6492 - 6503.

16. Hota, S.S.R.J.M.K., Removal of Powerline Interference from ECG Signal using FIR, IIR, DWT and NLMS Adaptive Filter. 2019 International Conference on Communication and Signal Processing (ICCSP), 4-6 April 2019.

17. Levkov , M., Ivanov , Daskalov , Christov , Dotsinsky Removal of power-line interference from the ECG: a review of the subtraction procedure. BioMedical Engineering OnLine, 23 Aug 2005

18. J.Mateoa, E.M.S.-M., J.L.Santos, A new method for removal of powerline interference in ECG and EEG recordings. Computers & Electrical Engineering, July 2015. 45: p. 235-248.

19. Ngoc Thang Bui, T.M.T.N., Sumin Park, Jaeyeop Choi, Thi Mai Thien Vo, Yeon-Hee Kang, Byung-Gak Kim, Junghwan Oh,, Design of a nearly linear-phase IIR filter and JPEG compression ECG signal in real- time system. Biomedical Signal Processing and Control, 2021. 67.

20. Tekale, A.S.B.R.G.S.V., Noise Analysis of ECG Signal Using Fast ICA. 2016 Conference on Advances in Signal Processing (CASP), 9-11 June 2016.

21. Nevi Jain, D.K.S., Denoising Baseline Wander Noise from Electrocardiogram Signal using Fast ICA with Multiple Adjustments. International Journal of Computer Applications, August 2014. 99(2): p. 0975 - 8887.

22. Rahul Kher*, Signal Processing Techniques for Removing Noise from ECG Signals. Journal of Biomedical Engineering and Research, March 12, 2019.

23. Pisani, B., Digital Filter Types in Delta-Sigma ADCs. May 2017.

24. Husain, K.M.Z., M.S.; Ul Hassan, S.; Hasbullah, S.; Mandala, S., Advances of ECG Sensors from Hardware, Software and Format Interoperability Perspectives. Electronics, 2021. 10(105).

25. Mohaddes, F.d.S., R.L.; Akbulut, F. P.; Zhou, Y.; Tanneeru, A.; Lobaton, E.; Lee, B.; Misra, V., A Pipeline for Adaptive Filtering and Transformation of Noisy Left-Arm ECG to Its Surrogate Chest Signal. Electronics 2020. 9.

26. Suranai Poungponsri, X.-H.Y., An adaptive filtering approach for electrocardiogram (ECG) signal noise reduction using neural networks. Neurocomputing,, 2013,. 117: p. 206-213.

27. Anuradha P, R.A., Rajkumar K,, Noise removal of ECG signals with adaptive filtering. Materials Today: Proceedings, 2021,.

28. Evangelin Dasan, I.P., A novel dimensionality reduction approach for ECG signal via convolutional denoising autoencoder with LSTM. Biomedical Signal Processing and Control, 2021. 63.

29. Baker, B., Delta-sigma ADCs in a nutshell. December 14, 2007.

30. Kumngern, M. A., N.; Khateb, F.; Kulej, T., 0.5 V Fifth-Order Butterworth Low-Pass Filter Using Multiple-Input OTA for ECG Applications. Sensors 2020. 20.

31. Husain, K.M.Z., M.S.; Ul Hassan, S.; Hasbullah, S.; Mandala, S., Advances of ECG Sensors from Hardware, Software and Format Interoperability Perspectives. Electronics 2021. 10.

32. Zompanti, A.S., A.; Grasso, S.; Pennazza, G.; Ferri, G.; Barile, G.; Chello, M.; Lusini, M.; Santonico, M., Development and Test of a Portable ECG Device with Dry Capacitive Electrodes and Driven Right Leg Circuit. Sensors 2021. 21.

33. Ngoc Thang Bui, T.H.V., Byung-Gak Kim, Junghwan Oh, Design of a Solar-Powered Portable ECG Device with Optimal Power Consumption and High Accuracy Measurement. Applied sciences, 2019. 9(10).

34. Plawiak, P., ECG signals (1000 fragments). 2017.

35. Instruments, T., ADS1293 24-bit, 3-ch, Low-Power Analog Front END (AFE) for ECG Applications. 17 Dec 2014.

36. Integrated, M., MAX86150 Integrated Photoplethysmogram and Electrocardiogram Bio-Sensor Module For Mobile Health, M. Integrated, Editor. 2018.

37. Christopher Beach, S.K., James Pope, Xenofon Fafoutis, Robert J. Piechocki3, Ian Craddock3, Alexander J. Casson, An Ultra Low Power Personalizable Wrist Worn ECG Monitor Integrated With IoT Infrastructure. 10 August 2018. 6: p. 44010 - 44021.

38. Hua, L.L.L.H.Y.Z.T., A Battery-Less Portable ECG Monitoring System With Wired Audio Transmission. IEEE Transactions on Biomedical Circuits and Systems, 20 June 2019. 13(7): p. 697 - 709.

39. Ming, J.L.G.D.W.W.H.W.Z., Design of a Real-Time ECG Filter for Portable Mobile Medical Systems. IEEE Access 26 October 2016. 5: p. 696 - 704.

40. Lab, T.M.P.V. Electrocardiogram Waveform. 2021 [cited 2021 May 15th]; Available from: https://www.medicine.mcgill.ca/physio/vlab/cardio/introECG.htm.

41. Bae, T.W.L., S.H.; Kwon, K.K., n Adaptive Median Filter Based on Sampling Rate for R-Peak Detection and Major-Arrhythmia Analysis. Sensors 2020. 20.

42. Zhang, D. W., S.; Li, F.; Wang, J.; Sangaiah, A. K.; Sheng, V. S.; Ding, X., An ECG Signal De-Noising Approach Based on Wavelet Energy and Sub-Band Smoothing Filter. Appl. Sci., 2019. 9.

43. Huang, Y.S., Y.; Gou, L.; Zou, Y., A Novel Wearable Flexible Dry Electrode Based on Cowhide for ECG Measurement. Biosensors 2021. 11.

Claims (8)

setting, by a filtering unit, parameters of a high-pass filter and a low-pass filter including an order and a movement delay interval of the filter for removing noise from the ECG signal;
extracting, by a sampling unit, an electrocardiogram signal using a plurality of different sampling frequencies; and
Comparing and analyzing before and after applying the high-pass filter and the low-pass filter to the extracted ECG signal by the analysis unit and the plurality of different sampling frequencies
A characteristic analysis method of an electrocardiogram signal comprising a. According to claim 1,
The step of the filtering unit setting parameters of the high-pass filter and the low-pass filter including the order and movement delay interval of the filter for removing the noise of the electrocardiogram signal,
setting parameters for an Infinite Impulse Response (IIR) filter and a symmetric Finite Impulse Response (FIR) filter, respectively;
The symmetric finite impulse response filter has an integer delay when the coefficients are odd,
The order of the infinite impulse response filter is set to be smaller than the coefficient of the symmetric finite impulse response filter, and only the infinite impulse response filter is applied to the high pass filter.
Method of characterization of electrocardiogram signals. The method of claim 1,
The step of the sampling unit extracting the electrocardiogram signal using a plurality of different sampling frequencies,
Generate the data about the ECG signal extracted with the sampling frequency from the BIT/MIH database as the original data,
Data on the ECG signal extracted using a plurality of different sampling frequencies are generated as comparison data in the MIT/BIT database.
Method of characterization of electrocardiogram signals. 4. The method of claim 3,
Comparing and analyzing the plurality of different sampling frequencies before and after the analysis unit applies the high-pass filter and the low-pass filter to the extracted ECG signal,
The criteria for comparing the original data and the comparison data before and after applying the high-pass filter and the low-pass filter include a signal-to-noise ratio, and P wave and QRS wave of the electrocardiogram signal for a plurality of different sampling frequencies , to compare the amplitude of the T wave
Method of characterization of electrocardiogram signals. a filtering unit including a high-pass filter and a low-pass filter applied by setting parameters including an order and a movement delay interval of a filter for removing noise from an electrocardiogram signal;
a sampling unit for extracting an electrocardiogram signal using a plurality of different sampling frequencies; and
An analysis unit that compares and analyzes the plurality of different sampling frequencies before and after applying the high-pass filter and the low-pass filter to the extracted ECG signal
An electrocardiogram signal feature analysis device comprising a. 6. The method of claim 5,
The filtering unit,
Parameters for an Infinite Impulse Response (IIR) filter and a symmetric Finite Impulse Response (FIR) filter are respectively set;
The symmetric finite impulse response filter has an integer delay when the coefficients are odd,
The order of the infinite impulse response filter is set to be smaller than the coefficient of the symmetric finite impulse response filter, and only the infinite impulse response filter is applied to the high-pass filter.
A device for analyzing the characteristics of an electrocardiogram signal. 6. The method of claim 5,
The sampling unit,
Generate the data about the ECG signal extracted with the sampling frequency from the BIT/MIH database as the original data,
Data on the ECG signal extracted using a plurality of different sampling frequencies are generated as comparison data in the MIT/BIT database.
A device for analyzing the characteristics of an electrocardiogram signal. 8. The method of claim 7,
The analysis unit,
The criteria for comparing the original data and the comparison data before and after applying the high-pass filter and the low-pass filter include a signal-to-noise ratio, and P wave and QRS wave of the electrocardiogram signal for a plurality of different sampling frequencies , to compare the amplitude of the T wave
A device for analyzing the characteristics of an electrocardiogram signal.

Images