KR102813203B1 - Method to remove noise from ECG signal - Google Patents

Abstract

The present invention divides a basic electrocardiogram signal into a low-frequency band and a high-frequency band by decomposition level through discrete wavelet transform, identifies detail coefficients by decomposition level, compares the absolute value of the detail coefficients identified by each decomposition level with a set threshold value, and makes the detail coefficients smaller than the threshold value '0', and then restores a first electrocardiogram signal with noise removed by using the detail coefficients by each decomposition level that are not '0' and the approximation coefficients by each decomposition level, and generates a second electrocardiogram signal by smoothing the first electrocardiogram signal using a Savitzky-Golay filter, and extracts a QRS section waveform of the basic electrocardiogram signal by using the QRS section identified in the second electrocardiogram signal and exchanges it with the QRS section waveform of the second electrocardiogram signal to generate a third electrocardiogram signal, and compares the P section waveform of the third electrocardiogram signal with the P section waveform of the basic electrocardiogram signal to determine distortion of the P section waveform of the third electrocardiogram signal, and corrects the distortion when distortion occurs to obtain a first result. The present invention relates to a method for removing noise from an electrocardiogram signal, which generates an electrocardiogram signal.

Description

{Method to remove noise from ECG signal}

The present invention relates to a method for removing noise included in a measured electrocardiogram signal.

Due to increasingly westernized eating habits, stress, and irregular lifestyles, modern people are more likely to be exposed to various heart diseases. In order to determine heart disease, the subject's electrocardiogram (ECG: ElectroCardioGram) signal must be measured through an electrocardiogram test, and the measured ECG signal must be analyzed.

The electrocardiogram signal is composed of a QRS complex, a P wave, and a T wave, and these are modified in amplitude, location, and interval depending on the type of heart disease, and the modification information of each of these waveforms is used as important information for diagnosing heart disease. Therefore, it is very important to reliably detect the components of the electrocardiogram signal in order to automatically diagnose heart disease.

However, the actual measured electrocardiogram signal has a problem in that it is difficult to detect the main waveform due to signal distortion such as baseline fluctuations caused by power noise, irregular breathing, and muscle noise.

Registered Patent No. 10-1983037 (Registration Date, 2019.05.22) Registered Patent No. 10-2216965 (Registration Date, 2021.02.10)

The present invention provides a noise removal method of an electrocardiogram signal, which enables the detection of the main waveform of an accurate electrocardiogram signal by changing the filter according to the noise level or type.

According to an embodiment of the present invention for achieving the above object, a noise removing method of an electrocardiogram signal comprises the steps of: receiving a basic electrocardiogram signal for a user; dividing the basic electrocardiogram signal into low-frequency components and high-frequency components by decomposition level through discrete wavelet transform; identifying a detail coefficient by the decomposition level; comparing the absolute value of the detail coefficient identified by each decomposition level with a set threshold value and making the detail coefficient smaller than the threshold value '0'; restoring a first electrocardiogram signal from which noise has been removed by using the detail coefficient by each decomposition level that is not '0' and the approximation coefficient by each decomposition level; generating a second electrocardiogram signal by smoothing the first electrocardiogram signal using a Savitzky-Golay filter; identifying the QRS segment from the second electrocardiogram signal; extracting a QRS segment waveform corresponding to the QRS segment from the basic electrocardiogram signal; generating a third electrocardiogram signal by exchanging the extracted QRS segment waveform with the waveform of the QRS segment of the second electrocardiogram signal; And it includes a step of comparing the P section waveform of the third electrocardiogram signal with the P section waveform of the basic electrocardiogram signal to determine distortion of the P section waveform of the third electrocardiogram signal, and correcting the distortion when distortion occurs to generate a first result electrocardiogram signal.

A method for removing noise from an electrocardiogram signal according to an embodiment of the present invention comprises the steps of: restoring an approximation coefficient of a low-frequency band obtained through the wavelet transform into a low-frequency signal through discrete wavelet transform; identifying a baseline from the low-frequency signal; and

The method may further include a step of generating a second result electrocardiogram signal by removing the baseline from the first result electrocardiogram signal.

In the step of generating the first result electrocardiogram signal, if the number of crossings between the P section waveform of the third electrocardiogram signal and the P section waveform of the basic electrocardiogram signal is less than a set number and the difference in values is greater than a set value, it is determined that there is distortion, and the P section waveform of the third electrocardiogram signal is filtered by reducing the window size of a second-order Savitzky-Golay filter by a set value to create a waveform approximate to the waveform of the P section of the basic electrocardiogram signal.

The above threshold value can be calculated through a setting program or arbitrarily set by the operator before measurement considering the measurement environment.

The waveform of the baseline identified from the above low-frequency signal may be smoothed by a Savitzky-Golay filter.

The step of generating the third electrocardiogram signal may include the step of inserting the extracted QRS segment waveform into the QRS segment of the second electrocardiogram signal; and the step of smoothing the connection portion of the signals before and after the QRS segment of the second electrocardiogram signal and the inserted QRS segment waveform.

The step of smoothing the above-mentioned connection portion may include the steps of: identifying a difference in the above-mentioned connection portion; determining a smoothing degree of the smoothing process according to a size of the difference; and applying the smoothing process to the above-mentioned connection portion by applying the determined smoothing degree.

The above basic electrocardiogram signal may be a signal obtained by removing baseline noise from a measured electrocardiogram signal measured for the user.

The present invention has the advantage of generating and providing an electrocardiogram signal by removing noise from an electrocardiogram signal through a discrete wavelet transform (DWT), adjusting a window size according to the level of the remaining noise, performing Savitzky-Golay filtering, and compensating for distortion of the QRS section waveform and the P waveform so as to facilitate detection of the main waveform.

FIG. 1 is a block diagram of a noise removal device for an electrocardiogram signal according to an embodiment of the present invention.
FIG. 2 is a flowchart of a method for removing noise from an electrocardiogram signal according to an embodiment of the present invention.
FIG. 3 is a flowchart of a noise removal method using discrete wavelet transform and thresholding according to an embodiment of the present invention.
FIG. 4 is a diagram for explaining a discrete wavelet transform (DWT) according to an embodiment of the present invention.
FIG. 5 is a signal waveform diagram for explaining a basic electrocardiogram signal according to an embodiment of the present invention.
FIG. 6 is a signal waveform diagram showing a discrete wavelet transformed signal and a signal filtered with a second-order Savitzky-Golay filter according to an embodiment of the present invention.
FIG. 7 is a signal waveform diagram for explaining waveform replacement of a QRS section according to an embodiment of the present invention.
FIG. 8 is a signal waveform diagram showing an example of applying a smooth filter to a waveform-replaced connection portion of a QRS section according to an embodiment of the present invention.
FIG. 9 is a signal waveform diagram for explaining a baseline signal according to an embodiment of the present invention.
FIG. 10 is a signal waveform diagram for explaining a baseline signal removal process according to an embodiment of the present invention.

Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Regardless of the reference numerals used in the drawings, identical or similar components will be given the same reference numerals and redundant descriptions thereof will be omitted. In addition, when describing embodiments disclosed in this specification, if it is determined that a detailed description of a related known technology may obscure the gist of the embodiments disclosed in this specification, the detailed description thereof will be omitted.

Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by the terms. The terms are used only to distinguish one component from another.

Singular expressions include plural expressions unless the context clearly indicates otherwise.

In this application, each step described may be performed regardless of the listed order, except in cases where it must be performed in the listed order due to a special causal relationship.

In this application, it should be understood that terms such as “comprises” or “have” are intended to specify the presence of a feature, number, step, operation, component, part or combination thereof described in the specification, but do not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

Hereinafter, the present invention will be described with reference to the attached drawings.

FIG. 1 is a block diagram of a noise removing device for an electrocardiogram signal according to an embodiment of the present invention. Referring to FIG. 1, a noise removing device (100) for an electrocardiogram signal according to an embodiment of the present invention includes a wavelet transform unit (101), a high and medium frequency noise removing unit (102), a smoothing unit (103), a QRS segment combining unit (104), a QRS extraction unit (105), a P segment processing unit (106), a baseline processing unit (107), a final signal generating unit (108), and an output unit (109).

The wavelet transform unit (101) receives a continuous basic electrocardiogram (ECG) signal (S1) for a set period of time (e.g., 2 seconds, 3 seconds, etc.) and performs a discrete wavelet transform on the received basic ECG signal (S1) to divide it into a mid-frequency band and higher band including a low-frequency band and a high-frequency band.

Here, the low-frequency band means a band below approximately 2 Hz that includes the baseline noise of the electrocardiogram, the high-frequency band means a band above approximately 40 Hz that includes EMG noise, and the mid-frequency band means a band in between. Depending on the level of the set wavelet transform and the sampling rate of the signal, the divided band varies, and the mid-frequency or higher band that can be removed also varies.

Referring to FIG. 4, an operation of an embodiment of a wavelet transform unit (101) is described. FIG. 4 is a diagram for explaining a discrete wavelet transform according to an embodiment of the present invention, and is an example of separating a frequency band into a third level. Of course, the wavelet transform unit (101) may be set lower than the third level or higher than the third level, and in order to remove a baseline, more levels must be passed through to extract the low-frequency band specified above from a signal than when removing noise above a mid-frequency, so the wavelet segmentation level executed to remove noise above a mid-frequency and the segmentation level for removing a baseline may be different.

The wavelet transform unit (101) calculates detail coefficients through high-pass filtering each time the frequency is divided by each level, and calculates approximation coefficients through low-pass filtering. At this time, the detail coefficients contain the high-frequency components of the signal of the corresponding level, and the approximation coefficients contain the low-frequency components. Since a downsampling process is performed each time the frequency is divided by each level, the lower the frequency band, the lower the resolution, and conversely, the higher the frequency band, the higher the resolution. At the first level, it is divided (decomposed) into detail coefficients (high-frequency components) and approximation coefficients (low-frequency components), at the second level, the low-frequency components at the first level are divided into low-frequency components and high-frequency components, and at the third level, the low-frequency components at the second level are divided into low-frequency components and high-frequency components.

Accordingly, the frequency components divided in the wavelet transform unit (101) are, assuming that the decomposed levels are 3 as in the example of Fig. 4, a low-frequency component divided in the third level and three different high-frequency components divided in the first to third levels. Here, among the three different high-frequency components, the frequency band of the high-frequency component divided in the first level is the highest, and the frequency band of the low-frequency component divided in the third level is the lowest.

The wavelet transform unit (101) provides detailed coefficients used for each level to the high- and medium-frequency noise removal unit (102). In addition, the wavelet transform unit (101) restores each divided frequency band provided at the request of the high- and medium-frequency noise removal unit (102) into an electrocardiogram signal (hereinafter referred to as the 'first electrocardiogram signal') from which noise above the medium-frequency band has been removed.

The high-frequency noise removing unit (102) removes noise in the high-frequency and higher bands. For example, the high-frequency noise removing unit (102) performs thresholding processing to compare the absolute value of the detailed coefficients with a set threshold value and make the detailed coefficients below the threshold value '0'. Making the detailed coefficients '0' means that the detailed coefficients at the corresponding level are not used when restoring (synthesizing) the discrete wavelet transform, which means that the high-frequency components divided at the corresponding level are not restored but removed as noise.

Meanwhile, the 'threshold value' above may be received from an external device or set in the high-frequency noise removal unit (102) of the noise removal device (100) or the threshold value setting unit (not shown).

The threshold value is a value set as a set percentage (e.g., 10%, 15%, 20%, etc.) of the peak value of the R waveform, and can be automatically calculated by the setting program or calculated based on a value provided by the user in advance. If the user provides a value, it may be when the measurement environment is unusual, such as the subject's skin type or surrounding environment.

When the high-frequency noise removal unit (102) completes threshold processing on the received basic electrocardiogram signal, it provides the wavelet transform unit (101) with detailed coefficients for each level or detailed coefficients processed with a '0' value, and instructs the unit to generate a first electrocardiogram signal through a restoration (combination) operation of discrete wavelet transform using the provided detailed coefficients.

The smoothing unit (103) includes a second-order Savitzky-Golay filter, receives a first electrocardiogram signal from the wavelet transform unit (101), determines a noise level included in the first electrocardiogram signal, sets a window size of the Savitzky-Golay filter that matches the noise level, and filters the first electrocardiogram signal with the set window size to generate a second electrocardiogram signal. The second electrocardiogram signal filtered with the second-order Savitzky-Golay filter becomes smooth by lowering and flattening the peak value of each waveform.

The QRS segment coupling unit (104) identifies the segment of each QRS group (hereinafter referred to as “QRS segment”) based on time in the second electrocardiogram signal, provides information (i.e., time information) on the QRS segment to the QRS extraction unit (105), receives the QRS waveform of the corresponding QRS segment to the QRS extraction unit (105), and generates a third electrocardiogram signal by replacing the received QRS waveform with the waveform of the corresponding QRS segment. The QRS waveform means a waveform of the QRS group, i.e., a waveform composed of a Q waveform, an R waveform, and an S waveform.

Here, if the received QRS waveform is replaced with the waveform of the corresponding QRS section, the continuity of the signal size may not be satisfied in the connection portion of the QRS signal that is replaced and inserted with the signal before and after the QRS section of the existing second electrocardiogram signal. This discontinuity feature may cause an error or incompleteness of the signal. Therefore, in the present invention, the connection portion may be modified so that the continuity is satisfied by applying a smoothing process.

The smoothing process applied to smooth the connected portion of the inserted QRS signal by replacing the signals before and after the QRS segment of the existing second electrocardiogram signal can be set differently depending on the degree of discontinuity of the connected portion.

For example, when the discontinuity difference of the connected portion is large, the window of the applied smoothing filter can be made relatively large to maximize correction of the discontinuity with a large difference, thereby enhancing continuity. On the other hand, when the discontinuity difference of the connected portion is small, the window of the applied smoothing filter can be made relatively small to sufficiently secure continuity while minimizing signal distortion caused by the smoothing filter.

The QRS extraction unit (105) extracts the QRS waveform of the basic electrocardiogram signal located in the information on the QRS segment received from the QRS segment coupling unit (104) and provides the extracted QRS waveform to the QRS segment coupling unit (104).

The P section processing unit (106) receives the third ECG signal, identifies the P section from the third ECG signal, and then compares (i.e., superimposes) the P section waveform of the third ECG signal corresponding to the identified P section with the P section waveform of the basic ECG signal to identify the number of crossings and the difference in signal size between the two waveforms. If the number of crossings between the two waveforms is large, the P section processing unit (106) determines that it is a normal P section waveform and uses the P section waveform of the third ECG signal as is, and if the number of crossings between the two waveforms is smaller than a set number and the difference in signal size is larger than the set size, it determines that distortion has occurred in the P section waveform and corrects the P section waveform of the third ECG signal.

In the above, the number of times can be set to 4 or 5 times, etc. In addition, the setting size is set to be larger than the average size difference between the P section waveform of the normal third electrocardiogram signal and the P section waveform of the basic electrocardiogram signal.

If the P section processing unit (106) determines that distortion has occurred in the P section waveform of the third electrocardiogram signal, it filters the distorted P section waveform by reducing the window size of the second-order Savitzky-Golay filter, thereby making it approximate the P section waveform of the basic electrocardiogram signal.

The electrocardiogram signal output from the P section processing unit (106) is either the same signal as the third electrocardiogram signal or an electrocardiogram signal with the distortion of the P section waveform compensated for. The electrocardiogram signal output from the P section processing unit (106) is collectively called the first result electrocardiogram signal.

The baseline processing unit (107) determines whether there is a baseline in the low-frequency signal generated by restoring only the low-frequency band in the wavelet transform unit (101), and if it is determined to be a baseline, smoothes the received low-frequency signal with a second-order Savitzky-Golay filter. At this time, the window size of the Savitzky-Golay filter is set to a sample size corresponding to approximately 1 second.

The final signal generation unit (108) receives the first result electrocardiogram signal of the P section processing unit (106) and the baseline signal output from the baseline processing unit (107), and generates the second result electrocardiogram signal by removing the baseline signal from the first result electrocardiogram signal. Here, in order to perform the work of extracting and removing the baseline, more decomposition levels must be passed through compared to the signal processing work of the mid-frequency or higher band, and thus a longer input signal is required. Therefore, in order to avoid delay in real-time waveform output, the first result electrocardiogram signal may be displayed on the screen before the second result electrocardiogram signal.

The output unit (109) outputs the second result electrocardiogram signal generated by the final signal generation unit (108) to the outside.

FIG. 2 is a flowchart of a noise removing method of an electrocardiogram signal according to an embodiment of the present invention. Referring to FIG. 2, a noise removing device (100) according to an embodiment of the present invention attempts to create an electrocardiogram signal, such as FIG. 5 (a), which is almost identical to a raw electrocardiogram signal, such as FIG. 5 (b), while having noise removed. That is, FIG. 5 (a) is an electrocardiogram signal to be finally created, and FIG. 5 (b) is a raw electrocardiogram signal. In addition, FIG. 5 (c) is a basic electrocardiogram signal obtained by bandpass filtering a raw electrocardiogram signal and removing a power line through a notch filter.

The noise removal device (100) receives a basic electrocardiogram signal as shown in (c) of Fig. 5 (S201). The noise removal device (100) can receive a raw electrocardiogram signal or a basic electrocardiogram signal through a bandpass and notch filter, depending on the user's selection.

The noise removal device (100) divides the received basic electrocardiogram signal into a plurality of frequency bands by dividing the signal from the first level to the set n levels through discrete wavelet transform (S202). The plurality of frequency bands are bands corresponding to a low frequency band to a high frequency band, and include a low frequency band, a mid frequency band, and a high frequency band.

The noise removal device (100) identifies the detailed coefficients for each decomposition level during discrete wavelet transform, compares the absolute value of the detailed coefficients for each decomposition level with a set threshold value, and performs a noise removal operation to make the detailed coefficients smaller than the threshold value '0' (S203).

The noise removal device (100) restores the electrocardiogram signal as in (a) of Fig. 6 by using the non-zero level-specific coefficients among the decomposition level-specific coefficients and the approximation coefficients for each decomposition level (S204). At this time, the restored electrocardiogram signal has noise in the band above the middle frequency removed, and is called the 'first electrocardiogram signal'.

The noise removal device (100) smoothes the peak value of the peak waveform of the first ECG signal using a second-order Savitzky-Golay filter to generate a second ECG signal as in (b) of Fig. 6 (S205). Here, when the first ECG signal and the second ECG signal are compared, it can be seen that the R peak value of the first ECG signal is lowered to the R peak value of the second ECG signal by filtering by the second-order Savitzky-Golay filter. This phenomenon is clearly shown in (c) of Fig. 6, and the R peak value of the second ECG signal of (c) of Fig. 6 is different from the R peak value of the first ECG signal.

In order to eliminate the difference from the basic electrocardiogram signal for the R peak value, the noise removal device (100) determines the time interval in which the QRS segment, i.e., the QRS group, is located in the second electrocardiogram signal illustrated in (b) of FIG. 7, extracts the QRS segment waveform located in the QRS segment (i.e., the time interval) determined above from the basic electrocardiogram signal illustrated in (a) of FIG. 7, and then replaces the extracted QRS segment waveform with the waveform of the corresponding QRS segment of the first electrocardiogram signal to generate a third electrocardiogram signal as illustrated in (c) of FIG. 7 (S206).

Here, as in (a) of Fig. 8, among the third ECG signals generated by replacing the extracted QRS waveform with the waveform of the corresponding QRS segment, the continuity of the signal size may not be satisfied in the connected portion of the QRS signal that is replaced with the signals before and after the QRS segment of the existing second ECG signal. In order to satisfy the continuity of the signal size, as in (b) of Fig. 8, the connected portion of the third ECG signal may be smoothed by applying a smoothing filter. At this time, the window of the smoothing filter may be set differently depending on the degree of discontinuity of the connected portion. For example, when the discontinuity difference of the connected portion is large, the window of the applied smoothing filter may be relatively large so as to correct the discontinuity with a large difference as much as possible to strengthen the continuity, and when the discontinuity difference of the connected portion is small, the window of the applied smoothing filter may be relatively small so as to sufficiently secure the continuity while minimizing signal distortion caused by the smoothing filter.

Then, the noise removal device (100) determines whether the P section waveform of the third electrocardiogram signal is distorted by filtering, and if so, performs an operation to compensate for the distortion (S207).

Specifically, the noise removal device (100) compares the P section waveform of the third ECG signal with the P section waveform of the basic ECG signal, and if it determines that the P section waveform of the third ECG signal is not distorted, it uses the third ECG signal as is as a first result ECG signal, and if it determines that distortion has occurred, it filters the P section waveform of the third ECG signal by reducing the window size of a second-order Savitzky-Golay filter to make it almost identical to the P section waveform of the basic ECG signal and uses it as a first result ECG signal.

The first result ECG signal is the signal that is displayed to the user in real time and is a signal that includes a baseline.

The baseline is included in the raw electrocardiogram signal as shown in (a) of Fig. 9, and is also included in the basic electrocardiogram signal from which basic noise has been removed using a bandpass filter and a notch filter as shown in (b) of Fig. 9, and is also included in the first electrocardiogram signal from which noise has been removed using thresholding using discrete wavelet transform as shown in (c) of Fig. 9.

The noise removal device (100) removes the baseline included in the first result electrocardiogram signal to create an electrocardiogram signal such as that in Fig. 5 (a) by performing a process of removing the baseline (S208 to S211).

When explaining the S208 process to the S211 process, the noise removal device (100) restores the low-frequency band divided in the S202 process into a low-frequency signal as in (a) of FIG. 10 through a discrete wavelet transform using an approximation coefficient of the last decomposition level (S208), identifies a baseline from the restored low-frequency signal (S209), smoothes the identified baseline using a second-order Savitzky-Golay filter as in (b) of FIG. 10 (S210), and removes the baseline included in the first result ECG signal using the smoothed baseline from the first result ECG signal to create a second result ECG signal as in (c) of FIG. 10 (S211).

And, the noise removal device (100) outputs the second result electrocardiogram signal on the screen or to the outside (S212). The second result electrocardiogram is the signal that is finally shown to the user and is the signal from which the baseline is finally removed from the first result electrocardiogram signal shown in real time.

FIG. 3 is a flowchart of a noise removal method using discrete wavelet transform and thresholding according to an embodiment of the present invention. Referring to FIG. 3, a noise removal device (100) identifies a peak value of a QRS signal, i.e., a QRS interval waveform, from a basic electrocardiogram signal (S301), and sets a set % as a threshold value in comparison to the identified peak value (S302). For example, if the peak value is '10' and the set % is 10%, the threshold value becomes '1'.

The noise removal device (100) performs discrete wavelet transform on the basic electrocardiogram signal received during the set sampling time. For example, if the set sampling time is 2 seconds, the basic electrocardiogram signal is received for 2 seconds (S303), and the first level of decomposition is performed on the basic electrocardiogram signal received for 2 seconds using the set wavelet function to separate low-frequency components and high-frequency components (S304).

The noise removal device (100) determines the first detailed coefficient used for the decomposition of the first level (S305) and compares the absolute value of the first detailed coefficient with a threshold value (S306). If the absolute value of the first detailed coefficient is greater than or equal to the threshold value, the noise removal device (100) leaves the first detailed coefficient as is, and if the absolute value of the first detailed coefficient is less than the threshold value, the first detailed coefficient is processed as a value of '0' (S308).

The noise removal device (100) terminates the decomposition when the decomposition reaches the set level (S309) and ends the noise removal through threshold processing (S310).

On the other hand, if the decomposition is not performed to the set level, the next level of decomposition is performed, the detailed coefficients of the next level are identified (S311), and then the detailed coefficients of the identified next level are compared with the threshold value (S312). If the absolute value of the detailed coefficients of the next level is less than the threshold value, it is processed as a value of '0', and if it is greater than the threshold value, the next decomposition operation is performed, and this process is repeated.

The threshold for each level is updated at the set time, i.e. every set sampling time.

The technical features disclosed in each embodiment of the present invention are not limited to that embodiment, and, unless they are mutually incompatible, the technical features disclosed in each embodiment may be combined and applied to different embodiments.

Therefore, each embodiment will focus on explaining each technical feature, but unless each technical feature is incompatible with each other, it can be applied in combination with each other.

The present invention is not limited to the above-described embodiments and the attached drawings, and various modifications and variations are possible from the viewpoint of those skilled in the art to which the present invention pertains. Therefore, the scope of the present invention should be determined not only by the claims of this specification but also by the equivalents of these claims.

100: Noise removal device 101: Wavelet transform unit
102: High and medium frequency noise removal section 103: Smoothing section
104: QRS segment junction 105: QRS extraction section
106: P section processing unit 107: Baseline processing unit
108: Final signal generation unit 109: Output unit

Claims (8)

A step of receiving a basic electrocardiogram signal for a user;
A step of dividing the above basic electrocardiogram signal into low frequency components and high frequency components by decomposition level through discrete wavelet transform;
A step of identifying detail coefficients for each of the above decomposition levels;
A step of comparing the absolute value of the detailed coefficient identified for each decomposition level with a set threshold value and making the detailed coefficient smaller than the threshold value '0';
A step of restoring a first electrocardiogram signal with noise removed by using the decomposition level-specific detailed coefficients that are not '0' and the approximation coefficients for each decomposition level;
A step of generating a second electrocardiogram signal by smoothing the first electrocardiogram signal using a Savitzky-Golay filter;
A step of identifying the QRS segment from the second electrocardiogram signal;
A step of extracting a QRS segment waveform corresponding to the QRS segment from the above basic electrocardiogram signal;
A step of generating a third electrocardiogram signal by exchanging the extracted QRS segment waveform with the QRS segment waveform of the second electrocardiogram signal; and
A step of comparing the P section waveform of the third electrocardiogram signal with the P section waveform of the basic electrocardiogram signal to determine distortion of the P section waveform of the third electrocardiogram signal and correcting the distortion when distortion is detected to generate a first result electrocardiogram signal,
A method for removing noise from electrocardiogram signals. In the first paragraph,
A step of restoring the approximation coefficients of the low-frequency band obtained through the above wavelet transform into a low-frequency signal through discrete wavelet transform;
a step of identifying a baseline from the above low frequency signal; and
further comprising the step of generating a second result electrocardiogram signal by removing the baseline from the first result electrocardiogram signal.
A method for removing noise from electrocardiogram signals. In the first paragraph,
The step of generating the first result electrocardiogram signal is
If the number of crossings between the P section waveform of the third electrocardiogram signal and the P section waveform of the basic electrocardiogram signal is less than the set number and the difference in values is greater than the set value, it is determined that there is distortion, and the P section waveform of the third electrocardiogram signal is filtered by reducing the window size of the second-order Savitzky-Golay filter by the set value to create a waveform approximating the P section waveform of the basic electrocardiogram signal.
A method for removing noise from electrocardiogram signals. In the first paragraph,
The above threshold value is calculated through a setting program or is arbitrarily set by the operator before measurement considering the measurement environment.
A method for removing noise from electrocardiogram signals. In the second paragraph,
The waveform of the baseline identified from the above low-frequency signal is smoothed by the Savitzky-Golay filter.
A method for removing noise from electrocardiogram signals. In the first paragraph,
The step of generating the third electrocardiogram signal is:
A step of inserting the extracted QRS segment waveform into the QRS segment of the second electrocardiogram signal; and
A step of smoothing the connection portion of the signal before and after the QRS segment of the second electrocardiogram signal and the inserted QRS segment waveform.
A method for removing noise from electrocardiogram signals. In Article 6,
The step of smoothing the above connection part is,
A step of identifying the difference in the above connecting portion;
A step of determining the smoothing degree of the smoothing process according to the size of the above difference; and
A step of applying the smoothing process to the connection portion by applying the above-determined smoothing degree.
A method for removing noise from electrocardiogram signals. In the first paragraph,
The above basic electrocardiogram signal is a signal from which the basic noise is removed using a bandpass filter and a notch filter from the measured electrocardiogram signal for the user.
A method for removing noise from electrocardiogram signals.