KR20030069586A - Method of ECG Analysis using Polynomial Approximation - Google Patents

Abstract

PURPOSE: An electrocardiogram analysis method is provided to observe and analyze the pattern of electrocardiogram by comparing polynomial approximations. CONSTITUTION: An electrocardiogram analysis method comprises a step(1,2) of detecting characteristics from an electrocardiogram signal by using a high frequency component; a step(3) of performing a polynomial approximation to an ST segment; a step(4) of performing a polynomial approximation and detecting the ST segment significantly influenced by the noise and distorted polynomial; and a step(5) of analyzing and diagnosing the pattern of electrocardiogram by analyzing forms of the electrocardiogram signals.

Description

Electrocardiogram analysis using polynomial approximation {Method of ECG Analysis using Polynomial Approximation}

The present invention relates to the diagnosis of heart disease through the analysis of electrocardiogram signals, and more particularly, to a method of analyzing each form using a polynomial approximated by approximating a P-wave, a T-wave, and an ST segment to a polynomial. .

According to the prior art, the analysis of electrocardiograms is to observe frequency components through mathematical transformation, to detect arrhythmia by QRS detection, to sag or rise of ST segment, and to diagnose disease by analyzing electrocardiogram type depends on doctor's judgment Doing. That is, the conventional ECG analysis is a technique limited to the detection of the presence of P waves, T waves, QRS and the level of the ST segment.

The present invention compares the polynomials approximated by approximating each part of the electrocardiogram to the polynomial, in order to solve the problems of the prior art as described above, the present invention provides a method for diagnosing heart disease by analyzing the shape of an electrocardiogram signal. The goal is to get a way to run a diagnostic through a program.

1 is a complete flow chart of the present invention

2 is an approximate polynomial of resting ECG and ST segment

3 is an approximate polynomial of ECG and ST segment during exercise

4 is an approximate polynomial of ECG and ST segment after exercise

Fig. 5 shows error accumulation for the removal of the distorted polynomial approximation.

6 is ST segment comparison using approximated polynomials

7 is ST segment pattern analysis using the approximated polynomial

To this end, the configuration of the present invention can be divided into feature point detection, polynomial approximation, distortion signal detection and removal, and pattern analysis using a high frequency component of an electrocardiogram. FIG. 1 shows an analysis process of analyzing a shape by approximating an ST segment to a polynomial. have.

As a method of approximating the P wave, T wave and ST segment of the ECG signal to the polynomial, a feature point using high frequency components in the QRS, T wave and ST segments is detected from the ECG signal (2), and the ST segment (4) approximating and detecting (4) ST segments and distorted polynomials with high noise effects, and analyzing and diagnosing electrocardiogram patterns of cardiac diseases by morphological analysis of ECG signals. In order to select the interval of the polynomial approximation, the accurate detection of R, S, and T-waves, the amount of noise in the ST-segment approximated by the polynomial, the included noise and the degree of distortion of the approximated polynomial It consists of an algorithm representing the correlation of.

This diagnostic method by polynomial approximation of ECG enables us to know the change of heart rate and ST shape over time, and can detect false polynomial approximation that contains a lot of noise by calculating error with ECG.

In Figures 2 to 4 (a) shows the electrocardiograms of the steady state (Fig. 2) and exercise (Fig. 3), after the exercise (Fig. 4), respectively. In each figure, 'o' shown in (a) consists of the detection of feature points (R, S and T-wave) using the high frequency components of ECG (1).

2 to 4 (b) are polynomial expressions between the characteristic points (S-wave and T-wave intervals) detected in the electrocardiogram state after the exercise (FIG. 3) and after the exercise (FIG. 3). The approximate to is shown and the steady state ST section consists of sufficient data to approximate the polynomial. The electrocardiogram (1) continuously measured from the patient changes its cycle according to the state in which the patient is sleeping, exercising or resting. In other words, when you exercise or relax, your heart rate increases shorter than when you're at rest, and when you sleep, your period becomes relatively long. As the ECG cycle changes according to the condition of the patient, it is difficult to compare simple data between the ECGs in each state. For example, the above-mentioned electrocardiogram (1) signal reads data at intervals of 3ms. If one cycle of a normal person's electrocardiogram is 0.8 seconds, about 267 data are represented. But when you're exercising, tense, or excited, about one-half of an ECG is about half a second, and about 167 data represent it. Therefore, in the case of using the approximated polynomial, the data intervals of the ECGs of the approximated states are appropriately adjusted to take the same number of data from the electrocardiograms of the steady state and the exercise state, so that the error of the two signals can be calculated by simple calculation.

Before calculating the error by calculating the error of the two signals, we determine how much noise is included in the ST segment and how much influence the included noise has on the polynomial approximation process. In other words, if the characteristics of the ST segment or ST segment that contain a lot of noise are not properly approximated to the polynomial, the accuracy of the result is inferior even though the pattern analysis is performed.

FIG. 5 shows an ST segment with high polynomial approximation distortion and an error level for excluding it from pattern analysis. The upper figure shows the original ST segment data and the error of the polynomial approximated by each beat. A and B parts at the top of the figure are excluded from the pattern analysis, and the actual ECG of the part is shown like the A and B parts at the bottom. This error calculation can detect and remove both a noisy ST-segment and a distorted polynomial.

FIG. 6 is a comparison of the steady state, exercise, and post-exercise. FIG. 6 (a) shows that the ST segment during exercise is lowered overall than the stable state by comparing five data after approximation by a polynomial. FIG. 6 (b) shows the similarity of the ST segment shapes. The shapes were compared with 25 data in the state where the center points of the ST segments were matched. This indicates that more deformation of the ST segment occurred after the movement.

7 is an error calculated by extracting an ST segment from an electrocardiogram of a predetermined period (120 beats) using the AA 'pattern and the BB' pattern as a reference ST segment, respectively. It can be seen that as the error of and the error with the BB 'pattern becomes smaller, the AA' pattern changes to the BB 'pattern. The AA 'pattern is a J-type ST-segment that is commonly found on normal ECG, and the BB' pattern is a strain-shaped ST-segment that is typical of myocardial ischemia. Therefore, the electrocardiogram used in the analysis is characterized by being able to detect the symptoms of myocardial ischemia over time.

As described above, according to the present invention, a method of analyzing each part of the ECG signal by using the polynomial approximated to the polynomial, thereby observing and diagnosing the morphological change of the ECG according to the heart disease. It also has the effect of reducing the burden on the doctor's ECG reading.

Claims (6)

A method for automatically analyzing and diagnosing electrocardiogram of cardiac disease patients, the method of approximating ECG signal to polynomial, the method of excluding noise and distorted polynomial approximation of ECG signal from the analysis, and comparing the approximated polynomial The method and the ECG analysis method using the polynomial approximation to diagnose the abnormality by examining the morphological similarity of the electrocardiogram of the heart disease subject compared with the polynomial representing the ECG signal. The electrocardiogram analysis using polynomial approximation to diagnose cardiac diseases causing heart diseases such as myocardial ischemia (myocardial insufficiency), myocardial infarction, angina pectoris, ventricular hypertrophy, atrial load, etc. Way. In the method of approximating the ECG signal of claim 1 to a polynomial, an ECG analysis method using polynomial approximation for diagnosing abnormalities in a heart disease subject by applying an approximation method including a Newton polynomial approximation method and a Bernstein polynomial approximation method. The method of excluding the noise signal of the electrocardiogram data of claim 1 is to calculate the error between the approximated polynomial and the electrocardiogram approximated to the polynomial, thereby applying the method of excluding the electrocardiogram and the distorted polynomial approximation containing a lot of noise from the analysis. Electrocardiogram analysis method using polynomial approximation to diagnose subject's abnormalities. Comparing the approximated polynomials of paragraph 1 with each other, the electrocardiogram analysis method using polynomial approximation for diagnosing abnormalities in a heart disease subject by applying a method of observing changes over time of an electrocardiogram. Electrocardiogram form that can be diagnosed by electrocardiogram such as myocardial ischemia and ventricular hypertrophy of the subject of the cardiac disease of paragraph 1 above, the electrocardiogram using polynomial approximation to diagnose abnormality by examining the polymorphic approximation of the electrocardiogram of paragraph 1 above Analytical Method.