KR20110097304A - Heart sound analysis method - Google Patents

Abstract

The present invention decomposes at least one of the components of the normal heart sound in a matching pursuit method, resynthesizes them into at least one specific frequency band, and then determines whether the heart sound is normal using the resynthesized information. To provide a heart sound analysis method. The heart sound analysis method includes a segmentation step, a matching purge decomposition step, and a band configuration and synthesis step. The segmentation step segments a specific period of the heart sound data into a plurality of samples. The matching pursuit decomposition step decomposes the plurality of sampled data by a matching pursuit method, and generates a plurality of atoms having a form of frequency, magnitude, and phase in this process. In the band construction and synthesis step, the plurality of atoms are classified and synthesized into a plurality of bands, and a time interval and a schizophrenic observation result of the second heart sound are output from the synthesized band, and the atoms of a specific section are synthesized. Outputs the result of calculating the physical distance difference (PRD) between the synthesized signal and the stored original signal.

Description

Heart sound analysis method

The present invention relates to heart sound analysis, and in particular, by using the characteristics of the unique frequency components of each heart sound, such as the first heart sound, second heart sound and heart noise, matching pursuit decomposition and identification of the heart sound signal in the time band The present invention relates to a heart sound analysis method for analyzing a state of a heart sound which is unknown in an original heart sound signal through a resynthesis process of a frequency band only.

Auscultation and echocardiography showed relatively accurate analysis when first introduced into the clinic. But with the precise diagnosis of the heart, such as Doppler echocardiography, cardiac nuclear medicine, nuclear magnetic resonance, and angiography, cardiac echocardiography is the study of other biological signals: electrocardiogram (ECG), electromyogram (EMG). It was not active compared to. However, as the interest in the ubiquitous healthcare system has recently increased, a health care terminal based on the heart sound has been developed, and the interest in the research on the heart sound has been increased with the introduction of the electronic stethoscope. Auscultation is not only a basic means of early diagnosis of heart disease, but also a primary diagnosis of many diseases. The analysis or classification of the heart sound signal divides the various states of the heart sound into normal and abnormal with noise, and in the abnormal state, it helps to determine the symptoms of heart disease.

Medical specialists can diagnose abnormal symptoms by using auscultation or heart severity in case of abnormal heart sound with heart murmur, but rather, whether the condition of the first and second heart sounds, which are the elements of heart sound, is normal It is often difficult to determine whether or not. For the above reasons, the analysis of the heart sound is close to the normal heart sound, but it is more important to analyze and determine whether the heart sound is normal for the ambiguous signal.

The technical problem to be solved by the present invention is to decompose at least one of the components of the normal heart sound in a matching pursuit method, resynthesizing them into at least one specific frequency band, and then using the resynthesized information. It is to provide a heart sound analysis method for determining whether the heart sound is normal.

The heart sound analysis method according to the present invention for achieving the above technical problem, comprises a segmentation step, matching purge decomposition step and band configuration and synthesis step. The segmentation step segments a specific period of the heart sound data into a plurality of samples. The matching pursuit decomposition step decomposes the plurality of sampled data by a matching pursuit method, and generates a plurality of atoms having a form of frequency, magnitude, and phase in this process. In the band construction and synthesis step, the plurality of atoms are classified and synthesized into a plurality of bands, and a time interval and a schizophrenic observation result of the second heart sound are output from the synthesized band, and the atoms of a specific section are synthesized. Outputs the result of calculating the physical distance difference (PRD) between the synthesized signal and the stored original signal.

According to the present invention, the conventional sound and heart-noise-centered analysis and classification methods are close to the normal heart, but it is difficult to determine whether the sound is normal or not for an ambiguous signal. There is an advantage to make a more accurate judgment.

1 shows a heart sound analysis method according to the present invention.
2 is an embodiment of the band configuration and synthesis step and expert opinion input step shown in FIG.
FIG. 3 embodies the image synthesis band construction step 201 and the PRD synthesis band construction step 207 shown in FIG.
FIG. 4 illustrates a row signal and seven synthesis bands D1 to D7 shown in FIG.
FIG. 5 shows an enlarged and inter-sample calculation of the cleaved sound 403 shown in FIG.
FIG. 6 illustrates a signal model of the second heart sound S2 in the 120 Hz to 170 Hz synthesis band D5.
7 shows a hypothetical graph for calculating a PRD error.

DETAILED DESCRIPTION In order to fully understand the present invention, the operational advantages of the present invention, and the objects achieved by the practice of the present invention, reference should be made to the accompanying drawings that describe exemplary embodiments of the present invention and the contents described in the accompanying drawings.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Like reference numerals in the drawings denote like elements.

Before describing the heart sound analysis method according to the present invention will be described for the heart sound.

Heart sounds are the sounds produced by the heart during one cycle of cardiac exercise. Cardiac movement can be defined as the repetitive movement of the heart's contraction and expansion, which is caused by the opening and closing of the valve due to the pressure difference inside the heart. The heart sounds include first heart sounds (S1), second heart sounds (S2), third heart sounds (S3), fourth heart sounds (S4), and various heart noises. ) And the like. In a healthy person, only the first heart sound (S1) and the second heart sound (S2) are auscultated, and the third heart sound (S3) and the fourth heart sound (S4) are recognized as normal heart sounds in children or adolescents, but are mainly interpreted pathologically in adults. do. In general, the frequency band of the heart sound is 20 Hz to 120 Hz for the first heart sound S1, 20 Hz to 170 Hz for the second heart sound S2, and the third heart sound S3 and the fourth heart sound S4 are It is known as 20 Hz to 70 Hz.

The first heart sound (S1) is the sum of the sounds heard when the mitral valve and the tricuspid valve are closed, which is the loudest among the heart sounds. The mitral and tricuspid valves are closed when ventricular systolic pressure increases rapidly, exceeding atrial pressure. After the aortic valve opens, the pulmonary valve opens immediately.

The occurrence of the second heart sound S2 means the end of the ventricular contractor, and the sound A2 heard when the aortic valve is closed and the sound P2 when the pulmonary artery plate is closed are combined. These meniscus gradually decreases the volume of blood from the ventricles and closes at a moment when the intraaortic and pulmonary artery pressures exceed the intraventricular pressure. The period of time between the closing of the aortic valve and the opening of the mitral valve is called isovolumic relaxation, at which time intraventricular pressure drops rapidly. When the intraventricular pressure drops to some extent, the intraventricular and atrial pressures are equal, opening the mitral and tricuspid valves.

In the present invention, since the first heart sound (S1) and the second heart sound (S2) of the second heart sound (S2) known to be more important for the examination of heart disease, the third heart sound (S3), the fourth heart sound (S4) And other murmurs.

The division of the second heart sound (S2) can be regarded as a kind of heart disease finding. The criterion for the separation of the second heart sound (S2) is that A2 (when the aortic valve is closed) and P2 (pulmonary artery plate), which are components of the second heart sound (S2), If it is within 30ms (milliseconds), it is normal and if it is over, it is considered to be divided. Therefore, there is a need to distinguish which PCG signal is a normal heart or a split heart sound of the second heart sound S2.

In addition, in normal human living systems, the aorta is closed and then the pulmonary artery is closed. Therefore, the second heart sound (S2) generated by the motion of the aorta and the pulmonary artery is a synthesis of the sound of P2 followed by A2. The general disruption of the second heart sound S2 means that the time interval of P2 following A2 becomes longer. In some cases, however, P2 occurs before A2, leading to a paradoxical split. This mitosis can also be seen as a kind of disease.

If the pulse is fast, even an expert may not easily distinguish the four heart sounds from the first heart sound (S1) to the fourth heart sound (S4) with auscultation. In particular, the shape of the second heart sound (S2) signal when the split and the shape of the signal generated by the interference of the second heart sound (S2) and the third heart sound (S3) has a similar aspect. The symptoms caused by the division of the second heart sound S2 and the symptoms caused by the third heart sound S3 may be the same, but may also be different. Therefore, it is important to distinguish whether the shape of the displayed signal is due to the component cleavage of the second heart sound (S2) or the interference of the third heart sound (S3) to the second heart sound (S2). Do.

Therefore, the object of the present invention is summarized as follows.

1. Provides a method for determining the presence or absence of splitting of the second heart sound (S2) and the time interval between A2 and P2.

2. It provides a method of determining the presence or absence of mitotic phenomenon of the second heart sound (S2).

3. It provides a method of distinguishing whether the division of the second heart sound (S2) is a self-disruption phenomenon or a phenomenon caused by the interference of the third heart sound (S3) on the second heart sound (S2).

The present invention is intended to more accurately determine the state of the heart sound that is not known only by the original heart sound signal by processing the heart sound analysis from the viewpoint of bio-signal analysis based on the time-frequency characteristics of the heart sound signal.

To this end, segment the heart sound of any one period reflecting the characteristics of the heart sound, synthesize the segmented heart sound and the decomposed heart sound into a certain frequency band, and analyze the state of the heart sound that was not known in the original heart sound from the synthesized signal. . In particular, the matching pursuit method is used to decompose the heart sounds.

In the sinusoidal model, the signal s [n] may be expressed in the form of a linear sum of sinusoidal components as shown in Equation 1 below.

,

,

는

번째 프레임에서

번째 진폭, 주파수, 위상을 나타낸다. 정현파 파라미터의 예측을 위한 매칭 퍼슈잇 방법은 오류상쇄(error concealment) 원리에 바탕을 둔다. 수학식 1에 표시된 신호에 매칭 퍼슈잇 방법을 적용하여, 신호의 스펙트럼의 크기 및 위상들을 수학식 2와 같이 구할 수 있다. here

,

,

Is

In the first frame

The second amplitude, frequency, and phase. The matching pushsuit method for the prediction of sinusoidal parameters is based on the error concealment principle. By applying the matching Pursuit method to the signal represented by Equation 1, the magnitude and phases of the spectrum of the signal may be obtained as shown in Equation 2.

및

And

Since it is a generally known technique, even if the above calculation process is not explained in detail, it will not cause a problem in implementing the present invention. However, the matching push method generally has higher frequency resolution than the spectral peak detection method used for parameter prediction, and shows superior performance compared to the 512 point fast fourier transform (FFT) method after windowing. It has the advantage of having a resolution similar to the method of taking an FFT.

1 shows a heart sound analysis method according to the present invention.

Referring to Figure 1, the heart sound analysis method, heart sound data input step 101, segmentation step 102, matching pursuit decomposition step 103, band composition and synthesis step 104, expert opinion input step 105 And a storage step 106.

Since the heart sound signal is an analog signal, it must be sampled to convert it to a digital signal. It is known that the observation can be accurate only by observing the heart sound signal up to about 2KHz, so sample at least 4KHz. The sampled data will increase, but the shorter the sampling period, the smaller the difference between the digital and analog signals. Therefore, the heart sound data received in the heart sound data input step 101 is preferably at least 4KHz or more sampled data.

, 마지막 샘플이

인

의 수열이 된다. The sampled heart sound data is a series of data having repetitive periods. In the segmentation step 102, any one period in which the heart sound state is most characterized is selected as the representative of the heart sound data, and the selected heart sound of any one period Segment the data back into multiple samples. For example, if the data segmented into n samples, the first sample

, The last sample is

sign

Becomes a sequence of.

는 매칭퍼슈잇분해단계(103)에서 매칭 퍼슈잇 알고리즘을 이용하여 분해 된다. 매칭 퍼슈잇은 반복(iterative) 처리과정을 수행하는데, 심음신호와 최대 내적(maximum inner product)을 가지는 아톰(atom)을 재귀적으로 빼줌으로써 신호를 분해한다. 분해 된 결과로서 생성되는 아톰(atom)은 다음과 같이 표시할 수 있다. Segmented data

Is decomposed using the matching pursuit algorithm in the matching pursuit decomposition step 103. The matching pursuit performs an iterative process, which decomposes the signal by recursively subtracting an atom having a heart sound signal and a maximum inner product. The atom generated as a result of decomposition can be expressed as follows.

Atom = [frequency, magnitude, phase]

In this atom, the frequency follows the sampling rate of the original signal that was decomposed. If m decomposition points were used to decompose data with a sampling rate of 8KHz by the matching push method, there would be a total of m frequency search intervals between 0 and 8kHz, and frequencies up to 4KHz by sampling theory. You can see the component information. Since each of the m frequency search intervals has one atom column, the atom column eventually has the number of decomposition points. Each atom column generated by a frequency search point is not used in all, but only atoms of a certain frequency band are used.

In the band composition and synthesis step 104, a plurality of bands (X _band1 (t) to X _band7 (t)) of frequency bands to be analyzed, respectively, are synthesized and synthesized to observe the time interval and schizophrenia of the second heart sound (S2). The result is generated and output, and the result of calculating the physical distance difference between the synthesized signal and the stored original signal after synthesizing the atoms of a specific section. Seven synthesized bands (X _band1 (t) to X _band7 (t)) are combined with the original signal (not shown) to output a total of eight images. These synthesized bands observe the relative magnitudes or separation of the signal from the original signal in each band. In these synthesized bands, three tests related to the second heart sound S2 are performed. Detailed configuration will be described later.

According to the three test results in the band composition and synthesis step 104, medical professionals input 105 findings for each test. The input findings are stored 106 in the database with the initial input data and the results output from the synthesis step 104. These stored results can then be used to be compared to the same state of mind for another person, or to a temporally different mind for the same person.

2 is an embodiment of the band configuration and synthesis step and expert opinion input step shown in FIG.

Referring to Figure 2, the band composition and synthesis step 104, the image synthesis band configuration step 201, band-specific synthesis and image output step 202, the second heart sound (S2) time interval calculation step 203, A second heart sound (S2) schizophrenic observation step 205, PRD synthesis band configuration step 207, band-by-band synthesis and error calculation step 208 and database and comparison step 209 is provided.

In the image synthesis band construction step 201, the heart sound data, which are decomposed by the matching push method and becomes the Atom string, is divided into seven bands classified into different frequency bands. In this step, the frequency components of the heart sound are 20 Hz to 120 Hz for the first heart sound S1, 20 Hz to 170 Hz for the second heart sound S2, and 20 Hz to 70 Hz for the third heart sound S3 and the fourth heart sound S4. Let's use. Each of the seven bands has a frequency band of 1000Hz to 2000Hz, 500Hz to 1000Hz, 250Hz to 500Hz, 170Hz to 250Hz, 120Hz to 170Hz, 70Hz to 120Hz, and 20Hz to 70Hz. The seven bands contained almost no frequency band of the normal heart sound from 2000Hz to 250Hz, so the frequency band was divided evenly by 1/2 toward the lower level. A section including the frequency component section of the normal heart sound is divided unevenly on the basis of 20 Hz, 70 Hz, 120 Hz, and 170 Hz. For details, refer to FIG. 3.

FIG. 3 embodies the image synthesis band construction step 201 and the PRD synthesis band construction step 207 shown in FIG.

The image synthesis band 301 of FIG. 3 represents seven synthesis bands Band1 to Band7 generated in the image synthesis band construction step 201 shown in FIG. 2, and each band includes 1000 Hz to 2000 Hz, 500 Hz to 1000 Hz, 250Hz ~ 500Hz, 170Hz ~ 250Hz, 120Hz ~ 170Hz, 70Hz ~ 120Hz, 20Hz ~ 70Hz. The PRD synthesis band 302 of FIG. 3 calculates the PRD value error of each of the two PRD synthesis bands PRD1 and PRD2 included in the PRD synthesis band construction step 207 shown in FIG. The process is illustrated.

,

,

,

,

,

,

의 이미지는 출력되어 전문가의 밴드별 이미지 검사를 위해 모니터링 된다. 출력되는 이미지는 도 4를 참조하면 된다. In the band-specific synthesis and image output step 202, atoms of the seven bands Band1 to Band7 synthesized are synthesized. 7 signals combined with the original signal

,

,

,

,

,

,

The image of the image is output and monitored for expert band-by-band image inspection. The output image may refer to FIG. 4.

FIG. 4 illustrates a row signal and seven synthesis bands D1 to D7 shown in FIG.

In reference numerals 401 and 403 of FIG. 4, it can be seen that the heart sound analysis method according to the present invention satisfies the presence or absence of cleavage of the second heart sound S2 and the time interval test of A2 and P2. In the band 402 of 120Hz to 170Hz, the relative size of A2 is larger than that of P2. By using this, it can be seen that P2 is a schizophrenic generated first when the front portion becomes smaller in the inspection of the second heart sound S2.

)에 해당하는 1000Hz~2000Hz 범위의 신호를 관찰한다. 도 4를 참조하면, 정상적인 제2심음(S2)을 나타내는 밴드(401)에서는 A2와 P2가 분열되지 않았지만, 분열된 제2심음(S2)을 나타내는 밴드(403)에서는 원본신호에서는 정확하게 볼 수 없었던 A2와 P2의 간격을 관찰할 수 있다. 의학 전문가는 제1밴드(

)의 출력된 이미지에서 A2의 시작점과 P2의 시작점 사이의 샘플을 계산하게 된다. 이미 설명한 바와 같이, 제2심음(S2)의 분열판단의 기준인 30ms를 임계점으로 제2심음(S2)의 분열유무 및 시간간격을 알 수 있다. 분열된 시간 간격의 계산은 (관찰된 샘플 수/샘플링 율)이 되는데 이를 도5에 도시하였다. In the second heart sound (S2) time interval calculation step 203, the first band (8) out of the eight bands of the image output.

Observe the signal in the range of 1000Hz to 2000Hz. Referring to FIG. 4, in the band 401 representing the normal second heart sound S2, A2 and P2 were not split, but in the band 403 representing the split second heart sound S2, the original signal could not be accurately seen. The interval between A2 and P2 can be observed. The medical professional is the first band (

In the output image of), we calculate the sample between the starting point of A2 and the starting point of P2. As described above, it is possible to know whether or not the second heart sound S2 is broken and a time interval based on a threshold point of 30 ms, which is a criterion for the split determination of the second heart sound S2. The calculation of the cleaved time interval is (number of samples observed / sampling rate), which is shown in FIG.

FIG. 5 shows an enlarged and inter-sample calculation of the cleaved sound 403 shown in FIG.

In FIG. 5, only the portion 403 shown in FIG. 4 is enlarged and an absolute value is taken. The number of samples is calculated between the start point of A2 and the start point of P2 so that the split time of S2 can be known.

If the time interval between A2 and P2 falls within 30 ms in the second heart sound (S2) cleavage time test, it can be inferred that there is any symptom if a normal or longer time interval has elapsed, and the medical professional inputs the findings (204) do.

)에 해당하는 120Hz~170Hz 범위의 신호를 관찰한다. 이에 대응되는 밴드는 도 4에서의 부재번호 402 및 404에 해당한다. 상기의 밴드에서는 제2심음(S2)의 성분 요소인 A2와 P2의 상대적 크기가 변했음을 관찰할 수 있는데, 이는 원본신호(row signal)에서 볼 수 없었던 부분이다. 원본신호(row signal)에서 제2심음(S2)의 요소인 A2와 P2는 정상적인 순서로 발생되었고, 상기 밴드에서 시간적으로 앞의 신호인 A2가 뒤따라오는 신호 P2보다 크다는 것을 볼 수 있다. 제5밴드(

)에서 A2와 P2의 변화된 크기를 보고 의학 전문가는 이에 대한 소견을 입력(206)한다. In the second heart sound (S2) schizophrenic observation step 205, the fifth band (8) among the eight bands outputted from the image (

Observe signal in the range of 120Hz ~ 170Hz. Corresponding bands correspond to reference numerals 402 and 404 in FIG. 4. In the above band, it can be observed that the relative magnitudes of A2 and P2, which are the component elements of the second heart sound S2, have changed, which is not seen in the row signal. In the original signal (row signal), the elements A2 and P2 of the second heart sound S2 are generated in a normal order, and it can be seen that A2, which is the previous signal in time in the band, is larger than the following signal P2. 5th band

See the changed magnitudes of A2 and P2, and the medical professional enters the findings (206).

FIG. 6 illustrates a signal model of the second heart sound S2 in the 120 Hz to 170 Hz synthesis band D5.

Like the model in the upper part of the model shown in Figure 6, if the front part is large it can be seen that A2 occurred earlier than P2. On the contrary, as in the lower model of the model shown in FIG. 6, if the rear part is large, it may be inferred that P2 is a mitosis that occurred before A2.

PRD synthesis band configuration step 207 is a process of configuring a band for calculating the percent root-mean-square difference (PRD) of the synthesized signal after the synthesis of the specific interval atoms. PRD is used to calculate the physical distance difference between the synthesized signal and the original signal, it can be expressed as shown in equation (3).

는 원본신호이고,

는 분해 후 다시 합성 된 신호 그리고

는 심음신호의 평균이다. PRD는 값이 작을수록 원본신호와 가깝다는 것을 의미한다. 합성밴드는 PRD합성밴드1과 PRD합성밴드2의 두 개로 이루어져 있고, 도 3에 도시된 부재번호 302에 이를 도시하였다. PRD합성밴드1은 20Hz~170Hz의 대역을 밴드로 한 것이고, PRD합성밴드2는 50Hz~170Hz의 대역을 밴드로 한 것이다. In equation (3)

Is the original signal,

The synthesized signal after decomposition and

Is the average of the heart sound signal. PRD means that the smaller the value, the closer to the original signal. The synthesis band is composed of two of PRD synthesis band 1 and PRD synthesis band 2, which is shown in part number 302 shown in FIG. PRD synthesis band 1 is a band of 20Hz ~ 170Hz band, PRD synthesis band 2 is a band of 50Hz ~ 170Hz band.

The band-specific synthesis and error calculation step 208 calculates the PRD values of the two-band PRD synthesis band 1 and the PRD synthesis band 2 generated in the PRD synthesis band construction step. The difference between the PRD values output from the two bands is calculated.

7 shows a hypothetical graph for calculating a PRD error.

Referring to FIG. 7, if the synthesized PRD1 in the 0Hz to 170Hz band is obtained as the synthesized PRD2 in the 50Hz to 170Hz band, an error between the two PRDs is obtained, and there is a first heartbeat (S1) and a second heartbeat (S2) component. It can be seen that the error value of the signal with the first heart sound S1, the second heart sound S2, and the third heart sound S3 is greater than the signal.

The frequency component section of the third heart sound S3 is 20 Hz to 70 Hz, and overlaps with 20 Hz to 120 Hz of the first heart sound S1 and 20 Hz to 170 Hz of the second heart sound S2. If the heart sound data is divided between the second heart sound S2, the PRD1 and PRD2 error values may cause the PRD difference as much as the information of the first heart sound S1 and the second heart sound S2 in the 20 Hz to 50 Hz band is lost. . On the other hand, if the heart sound data is in a state where the third heart sound S3 interferes with the second heart sound S2, the error values of PRD1 and PRD2 are only the first heart sound S1 and the second heart sound S2 in the 20 Hz to 50 Hz band. In addition, the PRD difference lost until the information of the third heart sound (S3) will occur.

Comparing with the database (209) step, the error value output through the PRD error calculation method is the PRD error average value and the first heart sound of the heart sound having only the first heart sound (S1) and the second heart sound (S2) components of the existing database ( A process of comparing the PRD error average values of the heart sounds having the components of S1) to the third heart sound S3 and determining which state is near. Based on the database and comparison process 209 above, the medical professional inputs findings 210.

In the present invention, as shown in Table 1 below, the heart sounds are largely normal (S1 + S2), S2 cleavage (S1 + A2P2), S3 or S4 interference (S1 + S2 + S3 or + S4), and heart noise (S1 + S2 +). murmur) divided the total of 90 experimental data into four categories.

Count
PRD average (%) error (%)
20 Hz to 170 Hz 50 Hz to 170 Hz normal 17  4.8783 7.2683  2.3900 S2 cleavage 9  4.6433  6.2259 1.5826 S3 or S4 Interference 26  4.6187 11.7297 7.1110 Heart murmur 38 15.6441 16.9260 1.2819

[Table 1] It is more important to find the state abnormality of the second heart sound (S2) than the heart noise because the various types of heart noise are combined into one. The PRD mean from 20Hz to 170Hz was similar in the 4% bands of normal, second heart (S2) split, third heart (S3) or fourth heart (S4). In the case of heart noise, most of the frequency component range appears in the range of 20Hz ~ 170Hz of the normal heart sound, and when synthesized according to the range of the normal heart sound, the physical distance from the original signal increases, so the PRD value is large. However, when looking at the error values of the two PRD bands, it can be seen that the heart sound of the interference state of the third heart sound S3 or the fourth heart sound S4 is at least 4.7% larger than the other states. This shows that the PRD value of the interfering state of the third heart sound S3 or the fourth heart sound S4 is greatly changed as the information of 20 Hz to 50 Hz is lost as in the PRD error value assumption previously mentioned in the present invention.

While the present invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiments. It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the scope of the present invention.

101: Heart sound data input (4kHz or higher sampling rate) 102: Segmentation
103: Matching Pursuit Decomposition 104: Band Composition and Synthesis
105: Enter expert opinion (3 items) 106: Save database
201: Image composition band composition 202: Band-specific composition and image output
203: 1000 Hz to 2000 Hz, S2 time interval calculation 204: Expert opinion input (1)
205: 120 Hz to 170 Hz, S2 mitosis observation 206: Expert opinion input (2)
207: PRD synthesis band configuration 208: band-specific synthesis and error calculation
209: Comparison with the database 210: Expert opinion (3)
301: image synthesis band 302: PRD synthesis band
401: 1000 Hz to 2000 Hz band of S2 normal heart sound
402: 120 Hz to 170 Hz band of S2 normal heart sound
403: 1000 Hz to 2000 Hz band of S2 cleavage
404: 120 Hz to 170 Hz band of S2 cleavage

Claims (11)

A segmentation step 102 of segmenting one specific period of the heart sound data into a plurality of samples;
A matching pursuit decomposition step 103 of decomposing the plurality of sampled data by a matching pursuit method and generating a plurality of atoms having a shape of frequency, magnitude, and phase in the process; And
The plurality of atoms are classified and synthesized into a plurality of bands, a time interval and a schizophrenic observation result of the second heart sound (S2) are generated and output from the synthesized band, and the synthesized signals are stored after synthesizing the atoms of a specific section. And a band configuration and synthesis step (104) for outputting a result of calculating a physical distance difference (PRD) from the original signal. The method of claim 1, wherein the band composition and synthesis step (104),
An image synthesis band construction step (201) of setting a plurality of synthesis bands using the atom;
A band-specific synthesis and image output step 202 of synthesizing corresponding atoms according to the set synthesis band and outputting the synthesized band signal and the original signal as an image;
A second heart sound (S2) time interval calculation step (203) of calculating a time interval of the second heart sound (S2) signal divided into two by selecting one band from the plurality of bands output as an image; And
A second heart sound (S2) schizophrenic observation step 205 of selecting another one of a plurality of bands output as an image and observing whether there is a mitosis in the second heart sound S2;
A PRD synthesis band construction step 207 of synthesizing at least two PRD synthesis bands by combining atoms of a specific section;
A band-by-band synthesis and error calculation step (208) for calculating the PRD values and PRD values of the at least two PRD synthesis bands; And
PRD error average value and first heart sound (S1) of the heart sound having only the first heart sound (S1) and the second heart sound (S2) components of the existing database, the error output through the band-specific synthesis and error calculation step 208, And a comparison database (209) for comparing the PRD error average values of the heart sounds having the components of the second heart sound (S2) and the third heart sound (S3) to determine which state is close. The method of claim 2, wherein the PRD,
Heart sound analysis method that calculates using the average of the original signal, the original signal after decomposition and re-synthesized signal and heart sound signal. The method of claim 2, wherein the plurality of composite bands,
Heartbeat analysis method with 7 bands of 1000Hz ~ 2000Hz, 500Hz ~ 1000Hz, 250Hz ~ 500Hz, 170Hz ~ 250Hz, 120Hz ~ 170Hz, 70Hz ~ 120Hz, 20Hz ~ 70Hz. The method of claim 4, wherein
In the second heart sound (S2) time interval calculation step 203, a signal of a frequency band of 1000 Hz to 2000 Hz is used.
In the second heart sound (S2) schizophrenia observation step 205, a signal of a frequency band of 120 Hz to 170 Hz is used. The method of claim 2,
In the second heart sound (S2) schizophrenic observation step 203, the sound analysis that compares the relative magnitude of the sound (A2) heard when the aortic plate included in the second heart sound (S2) is closed and the sound (P2) heard when the pulmonary artery plate is closed Way. According to claim 2, The second heart sound (S2) schizophrenic observation step 203,
If the second heart sound (S2) is divided, the heart sound analysis method further determines whether the cleavage is inherent division of the second heart sound (2) or by the interference by the third heart sound (S3). The method of claim 7, wherein
Heart sound analysis method to calculate the error between the PRD of two synthetic bands of 20Hz ~ 170Hz, 50Hz ~ 170Hz, in order to determine whether it is due to inherent division and interference. The method of claim 1, wherein the heart sound data is sampled data having a sampling rate between 4 KHz and 8 KHz. The method of claim 1,
And an expert opinion input step (105) for inputting the judgment result by judging the result output from the band composition and synthesis step (104). The method of claim 10,
And a step of storing the result output from the band construction and synthesis step and the determination result in a database.

Images