KR101511204B1 - Unusual waveform detection method in ECG signal based on Refractory Period - Google Patents

Abstract

The present invention relates to a method to detect an unusual waveform from an arrhythmia electrocardiogram (ECG) signal based on a refractory period. The method includes the following steps of: (a) preprocessing an ECG signal; (b) extracting an R peak candidate from the ECG signal; (c) calculating a refractory period (RP) based on the kurtosis and potential of the R peak candidate; (d) determining whether a potential higher than the R peak candidate exists in the refractory period or not; (e) determining a current candidate peak as an R peak when a potential higher than the R peak candidate does not exist in the refractory period, or extracting the higher potential as a new R peak candidate when the higher potential exists, and repeating the steps from the step (a) to (C); and (f) detecting a waveform including the current R peak as an unusual waveform when the potential and kurtosis values of the determined R peak do not exist within a preset threshold value at the same time. The provided method to detect an unusual waveform of an ECG signal: is provided based on a refractory period; provides a stable and high detection rate in a simple way; improves a computing speed; can be applied to various kinds of applications for a real-time processing; and efficiently improves the efficiency of time and costs for the detection.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an ECG signal based on refractory period,

The present invention relates to a method of detecting an R wave in an electrocardiogram signal, and more particularly, to a method of detecting R wave using an adaptive reflex in an ECG signal applied with a reflexive period that reflects the kurtosis and potential of a candidate R- ≪ / RTI >

As domestic economic growth and living environment have been westernized recently, heart disease has become one of the major causes of death in Korea, along with cancer and cerebrovascular disease. Typical types of heart disease include coronary artery disease (CAD), arrhythmia, and heart failure. The most important signal in the diagnosis and prognosis of these heart diseases is the electrocardiogram (ECG) signal generated from the activity of the heart, and is particularly effective in the diagnosis of arrhythmia due to electrical stimulation of the heart and conduction disturbance.

In addition, early detection of heart disease requires long time measurements and accurate readings. In addition, due to the development of high-speed communication network and ubiquitous era, development of U-healthcare system capable of receiving services such as disease prevention and management anytime and anywhere is actively being carried out.

Signs such as arrhythmia in heart disease are rarely found in everyday life, so measurements and storage over 24 hours are needed. Therefore, data compression is also needed to store large amounts of data. However, distortion of electrocardiogram signal during restoration of compressed data may cause errors in health readings of the examinee. Also, it is necessary to analyze a large amount of data due to the characteristics of electrocardiogram (ECG) signals, which can not diagnose heart disease with only a small amount of data signals.

On the other hand, most of the arrhythmia ECG signals have an average waveform of about 10%. Therefore, over 90% of the electrocardiogram signal samples are given only reduced waveforms to medical staff who need to observe and analyze the electrocardiogram signal for a long time, which is a great time and cost effect. Various methods such as moving window integration, Hilbert transform, and wavelet transform have been studied in the waveform extraction algorithm for ECG diagnosis, and the waveform extraction rate is also high, but it is difficult to implement the system because of the large amount of calculation.

Distortion of the raw ECG signal can lead to errors in the health readings of the examinee. Also, it is necessary to analyze a large amount of data due to the characteristics of electrocardiogram (ECG) signals, which can not diagnose heart disease with only a small amount of data signals.

Korean Patent No. 10-1309309 Korean Patent No. 10-1029386

A problem to be solved by the present invention is to provide a method of detecting a specific waveform of an arrhythmia electrocardiogram signal based on a refractory period, in which at least 90% of a sample of an electrocardiogram signal is given a reduced specific waveform And a method of detecting a specific waveform of an electrocardiogram signal which is very efficient in terms of time and cost.

According to an aspect of the present invention, there is provided an electrocardiogram (ECG) signal processing method comprising the steps of: (a) preprocessing an ECG signal; (b) extracting R vertex candidates from the electrocardiogram signal; (c) calculating a refactoring period (RP) based on the kurtosis and the potential for the R vertex candidate; (d) determining whether or not there is a potential higher than the R vertex candidate in the refractory period; (e) if there is no potential higher than the R vertex candidate in the refractory period, the current candidate vertex is determined as the R vertex, and if there is a high potential, the high potential is extracted as a new R vertex candidate, Repeating; And (f) detecting the waveform including the current R vertex as a specific waveform when the determined potential value and the kurtosis value of the R vertex are not simultaneously within a preset threshold value range.

Here, in the step (f)

The threshold value range may be,

Where m _v and σ _v are the mean and standard deviation of the R-peaks not included in the previous singular waveform, m _k ? And? _K are the mean and standard deviation of the R-peaks not included in the previous singular waveform, And k ₁ ? And k ₂ are the threshold values of the potential and the kurtosis, respectively.

Preferably, the step (b) may be a step of detecting a point that is 20% or more of a previously detected R vertex potential with the R vertex candidate.

add,

The non-

는 기준 불응기이고,

는 후보 R-정점의 첨도이고,

는 이전 R-정점의 첨도이며,

는 후보 R-정점의 절대값 전위이고,

는 이전 R-정점의 절대값 전위이다.) 와 같은 식으로 산출되는 것이 바람직하다.(Where,

Is a reference refractory period,

Is the kurtosis of the candidate R-vertex,

Is the kurtosis of the previous R-vertex,

Is the absolute value potential of the candidate R-vertex,

Is the absolute value potential of the previous R-vertex).

It is preferable that the reference refractor is 35% of the PR interval, and the kurtosis of the R vertex candidate is < RTI ID = 0.0 >

Is calculated in the same manner as the above.

The present invention provides a method for detecting a specific waveform of an arrhythmia electrocardiogram signal based on a refractory period, provides a stable and high detection rate by a simple method, can further improve a calculation speed, The present invention provides a specific waveform detection method that can be applied to various applications, as well as a method of detecting a specific waveform of an electrocardiogram signal that can effectively improve time and cost efficiency in detection.

FIG. 1 is a schematic diagram showing characteristic points of an electrocardiographic waveform reflecting a cardiac electrical activation phase,
2 is a graph showing the cell membrane action potential of myocardial cells,
3 is a graph showing the R wave reflex of the MIT-BIH 101 record,
4 is a graph comparing waveforms of electrocardiographic signals and cell membrane activity potentials,
5 is a flowchart illustrating a specific waveform detection method in an arrhythmia ECG signal based on a non-adaptive period according to an exemplary embodiment of the present invention,
FIG. 6 is a graph showing the kurtosis of the R vertices detected by the specific waveform detection method in the arrhythmia ECG signal based on the refractory period according to the embodiment of the present invention,
7 is a specific waveform detection result for the MIT-BIH ADB 101 records,
8 is a specific waveform detection result for the MIT-BIH ADB 108 records,
FIG. 9 shows the result of detection of the specific waveform for the MIT-BIH ADB 208 records.

BRIEF DESCRIPTION OF THE DRAWINGS The advantages and features of the present invention, and how to accomplish it, will be described with reference to the embodiments described in detail below with reference to the accompanying drawings. However, the present invention is not limited to the embodiments described herein but may be embodied in other forms. The embodiments are provided so that those skilled in the art can easily carry out the technical idea of the present invention to those skilled in the art.

In the drawings, embodiments of the present invention are not limited to the specific forms shown and are exaggerated for clarity. Also, the same reference numerals denote the same components throughout the specification.

The expression "and / or" is used herein to mean including at least one of the elements listed before and after. Also, singular forms include plural forms unless the context clearly dictates otherwise. Also, components, steps, operations and elements referred to in the specification as " comprises "or" comprising " refer to the presence or addition of one or more other components, steps, operations, elements, and / or devices.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.

5 is a flowchart illustrating a specific waveform detection method in an arrhythmia ECG signal based on a non-adaptive period according to an exemplary embodiment of the present invention. As shown in FIG. 5, the method for detecting a specific waveform in an electrocardiogram signal according to an embodiment of the present invention includes: (a) preprocessing an ECG signal (S100); (b) extracting R vertex candidates from the electrocardiogram signal (S200); (c) calculating a refactoring period (RP) based on the kurtosis and the potential for the R vertex candidate (S300); (d) determining whether there is a potential higher than the R vertex candidate in the refractory period (S400); (e) if there is no potential higher than the R vertex candidate in the reflexive period, the current candidate vertex is determined as the R vertex (S450); if there is a high potential, the high potential is extracted as a new R vertex candidate; Repeating steps (S400); And (f) detecting the waveform including the current R vertex as a specific waveform when the determined potential value and the kurtosis value of the R peak are not simultaneously within a preset threshold value range (S500) (S600) .

As described above, the present invention proposes a method of detecting a specific waveform of an arrhythmia electrocardiogram signal based on a refractory period. Most of the arrhythmia electrocardiogram signals have an average waveform of about 10%, so long-term electrocardiographic signals must be observed and analyzed The present invention provides a method of detecting a specific waveform of an electrocardiogram signal that can provide a great effect in terms of time and cost by providing only a reduced waveform of at least 90% of the electrocardiogram signal samples to a medical staff.

Also, the method of detecting a specific waveform of an electrocardiogram signal according to the present invention detects an R-peak using a characteristic of an R-wave and a variable refractory, and detects a potential of R-peaks not included in a specific waveform with respect to the detected R- And the mean and standard deviation of the kurtosis, the algorithm was applied to all records in the MIT-BIH arrhythmia database and showed an average compression rate of 90% or more.

Electrocardiogram signal and Refusal

The circulation of the heart is made because the myocardium repeats contraction and expansion. Whenever the heart beats, a weak electricity is generated, which causes current to flow in the body and this electric current forms a potential on the body surface. Electrocardiogram (ECG) is a small electric potential change induced by a certain method on the appropriate part of the body surface.

1 is a schematic diagram showing characteristic points of an electrocardiographic waveform that reflects an electrical activation phase of a heart. As shown in FIG. 1, the characteristic points of the ECG waveform are basically composed of P, Q, R, S, and T waves. P wave occurs at the time of atrial depolarization, QRS group shows ventricular depolarization, and T wave indicates ventricular repolarization. Atrial depolarization begins in the vicinity of the superior nodal region and proceeds from right to left across the atrium. Thus, the first part of the P wave represents the depolarization of the right atrium, and the posterior part of the P wave represents depolarization of the left atrium.

The QRS group represents the ventricular depolarization state. Ventricular depolarization begins at the left side of the interventricular septum near the AV junction and proceeds from left to right across the interventricular septum. The normal R wave has the highest and sharpest waveform in the QRS complex, Q wave represents ventricular septal depolarization, and the rest of the QRS complex is synchronous. It represents the depolarization of the right ventricle. The PR interval does not exceed 0.2 seconds in normal cases, within 0.12 seconds in the QRS group, and 0.4 seconds in the QT interval. The normal R wave is defined as the first recorded upward wave, Q wave is defined as the downward wave recorded before the R wave, and the S wave is defined as the downward wave recorded after the R wave.

It can be grasped by key elements such as duration, interval with neighboring wave, segment with neighboring wave, amplitude of wave, Especially, R wave of electrocardiogram signal is the largest signal, and it can be regarded as representative point. Since various feature points are detected based on this point, much effort has been made to develop an algorithm for improving R wave detection performance.

2 is a graph showing the cell membrane action potential of myocardial cells. As shown in Fig. 2, myocardium or nerves lose excitability for a certain period of time once they are excited, and they are not excited even if they are stimulated. If the ventricle contracts due to refusal of the myocardium, when the following stimulus comes, the ventricle is in the reflex phase and does not contract and becomes the resting phase. It divides into absolute refractory not responding to strong stimulus and relative refractory only to strong stimulus.

3 is a graph showing the R wave reflex of the MIT-BIH 101 record. As shown in FIG. 3, when the sample frequency of the electrocardiogram signal is 360 Hz, the RR interval is 875 ms for a normal electrocardiogram signal and 88 ms is about 10% of an RR interval for a QRS group. The time to Q-wave and R-peak potential is 44ms, which is about 5% of the RR interval. The absolute reflex termination time from the R-peak potential is 217ms, which is 25% of the RR interval. Which is about 175 ms.

4 is a graph comparing waveforms of electrocardiographic signals with cell membrane activity potentials. As shown in FIG. 4, when the QRS complex is started, it reaches the absolute refractory period for a certain period of time and enters the relative refractory period thereafter. As shown in Fig. 4, when the activity potential of myocardial cells is divided into stages, the QRS complex belongs to the first stage and the T wave belongs to the third stage. Here, the RR interval necessarily has time to the minimum absolute reflex period or relative reflex period .

However, the R wave occurs at the time of ventricular depolarization, and its shape differs from the normal waveform due to various factors (conduction, generation disorder). These obstacles affect the cycle from ventricular repolarization to depolarization, since the R wave is depolarized

The refractory time, which is the time that does not respond to other stimuli, changes. Accurate calculation of the variable refractor can therefore greatly improve the performance of waveform detection.

Since the arrhythmia ECG signal has an average waveform of about 10%, it is very important to detect a specific waveform with more than 90% of the ECG signal samples compressed when a long time ECG signal is observed and analyzed. In the embodiment of the present invention, a method of detecting a specific waveform of an arrhythmia ECG signal based on a variable refractory period is proposed. The proposed detection method uses a potential of a candidate R-peak, a variable refractory period of the myocardium, The exact R-peak is detected, and a specific waveform is detected using the mean and standard deviation of the potential and the kurtosis of the R-peaks not included in the specific waveform for the detected R-peak. The performance of the algorithm according to the proposed method depends on the accuracy of R-peak detection.

There are real-time QRS group detection algorithms proposed by Pan and tomkins, which are representative methods of R-peak detection. Although this algorithm shows good results for normal waveforms, it requires a large amount of computation to detect R waves, and has a significantly lower R wave detection rate in arrhythmia signals with many specific waveforms. The R-peak detection method using the adaptive reflex of the proposed algorithm has a detection rate of 99% or more and a small amount of computation, which is a real-time processing method.

In the embodiment of the present invention, R-peak detection is performed using an R-peak detection method using an adaptive reflex. FIG. 5 is a block diagram of an algorithm for detecting a specific waveform of an electrocardiogram signal according to an embodiment of the present invention, and comprises a preprocessing, candidate R-peak detection, refractory calculation, R-peak detection, .

Step 1: Perform preprocessing of ECG signal. Morphology pair method is used for preprocessing such as baseline smoothing (S100).

Step 2: Extract relatively high dislocation points as R-peak candidates. The candidate R-peak is extracted from the peak that is 20% or more of the detected R-peak potential (S200).

Step 3: Calculate the refractory RP based on the kurtosis and dislocation for the candidate R-peak extracted in step 2. The RP for the candidate R-peak is calculated in Equation (1) (S300)

Where RP _ref is 35% of the previous RR interval as a group based refractory, K _{R -? Peak} is the kurtosis of candidate R-peak, _peak K _{R _ _ prev} is the kurtosis of the previous R-peak, the candidate VR_peak R- V _{R _ peak _ prev} is the potential of the previous R-peak. Also, the kurtosis indicates the curvature, and the larger the value is, the sharpness of the waveform becomes. That is, the kurtosis can be obtained by using the potential gradient of the candidate R-peak.

As shown in FIG. 6, when the sample interval is 2.778 msec (sample frequency is 360 Hz) with respect to the upward and downward waveforms, the kurtosis of the candidate peak with reference to the vicinity of the 1/2 potential of the R-peak potential is expressed by Equation 2 same.

Step 4: Detect the R-peak using the high potential of the R-wave in the non-resonant mode. If there is no potential higher than that of the R-peak candidate, the current candidate peak is determined as the R-peak, and if there is a higher potential, the current candidate peak is discarded and the peak having the higher potential becomes the new candidate peak (S400).

Step 5: Detect the specific waveform according to the potential and the kurtosis value of the detected R-peak. And the current R-peak voltage value (V _{R _ peak)} and the Kurtosis value (K _{R_peak)} the formula 3 and the R-peak of the specific waveform does not meet the Equation 4] at the same time, specific to beat, including this (S500, S600).

Where m _v and σ _v are the mean and standard deviation of the R-peaks not included in the previous singular waveform, and m _k ? And? Σ _k are the mean and standard deviation of the kurtosis of the R-peaks not included in the previous singular waveform . And k ₁ ? And k ₂ are thresholds of potential and kurtosis, respectively. [Equation 3] and [Equation 4] are proposed on the assumption that the distribution of the potential and the kurtosis of the electrocardiogram obtained over a long period of time follows a normal distribution. Repeat steps 3 through 5 for the new R-peak.

3. Experiments and Considerations

To evaluate the performance of the proposed algorithm, MIT-BIH ADB, a standard DB used in electrocardiogram signal processing research, was used. Signals included in all records in the DB are the signals sampled at a sampling frequency of 360 Hz for 30 minutes. And records 102 and 104 are read 5 signals, and all other records are read 2 signals. The PC environment used in the experiments was Intel i3 550 CPU, 4G DDR3 RAM, Windows XP Professional operating system, and MATLAB version7.8 32bit simulation. The performance of detecting the specific waveform of the proposed algorithm is confirmed by using the compression ratio (CR) of the following [Equation 5].

Here, Total beats is the total number of beats included in 23 records of the MIT-BIH ADB, and Unusual beats is the total number of beat of the specific waveform detected by the proposed algorithm.

[Table 1] shows the experimental results of the proposed algorithm for 23 records of MIT-BIH ADB. As shown in the table, the proposed algorithm shows an average compression ratio of 90%. k ₁ ? and k ₂ were determined according to the standard deviation of the potential and the kurtosis. The experimental results in [Table 1] show the results when k ₁ ? And k ₂ are 1.65.

FIGS. 7, 8 and 9 show the detection results of the specific waveforms for the MIT-BIH ADB 101, 108, and 208 records. As shown in the figure, it can be seen that the proposed algorithm correctly detects the singular waveform.

As described above, the present invention proposes a detection algorithm as a method of detecting a specific waveform of an arrhythmia ECG signal based on a refractory period. First, R-peaks were detected on the basis of the variable imperfection, and specific waveforms were detected using the mean and standard deviation of the R-peaks of the R-peaks not included in the specific waveform for the detected R-peaks. Applying the proposed algorithm to all records in the MIT-BIH arrhythmia database, the average data compression rate was 90% or more. Therefore, more than 90% of the electrocardiogram signal samples are provided to the medical staff who need to observe and analyze the electrocardiogram signal over a long period of time.

While the invention has been shown and described with respect to the specific embodiments thereof, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined by the appended claims. Anyone with it will know easily.

Claims (6)

(a) preprocessing an electrocardiogram (ECG) signal;
(b) extracting R vertex candidates from the electrocardiogram signal;
(c) calculating a refactoring period (RP) based on the kurtosis and the potential for the R vertex candidate;
(d) determining whether or not there is a potential higher than the R vertex candidate in the refractory period;
(e) if there is no potential higher than the R vertex candidate in the refractory period, the current candidate vertex is determined as the R vertex, and if there is a high potential, the high potential is extracted as a new R vertex candidate, Repeating; And
(f) detecting the waveform including the current R vertex as a specific waveform when the determined potential value and the kurtosis value of the R vertex are not simultaneously within a preset threshold value range. Detection of specific waveforms in arrhythmia electrocardiogram signals.
The method according to claim 1,
In the step (f)
The threshold value range may be,

Where m _v and σ _v are the mean and standard deviation of the R-peaks not included in the previous singular waveform, m _k ? And? _K are the mean and standard deviation of the R-peaks not included in the previous singular waveform, And k ₁ ? And k ₂ are the thresholds of potential and kurtosis, respectively.)
And detecting the abnormal waveform in the arrhythmia electrocardiogram signal based on the refractory period.
3. The method of claim 2,
The step (b)
Detecting a point that is 20% or more of a previously detected R vertex potential as the R vertex candidate.
3. The method of claim 2,
The non-

(Where,

Is a reference refractory period,

Is the kurtosis of the candidate R-vertex,

Is the kurtosis of the previous R-vertex,

Is the absolute value potential of the candidate R-vertex,

Is the absolute value potential of the previous R-vertex.)
And calculating an R-wave using the adaptive reflex in the electrocardiogram signal.
5. The method of claim 4,
Wherein the reference refractor is 35% of the previous PR interval.
The method according to claim 1,
The kurtosis of the R vertex candidates may be expressed as:

And calculating an R-wave using the adaptive reflex in the electrocardiogram signal.

Images