JP5190954B2 - Heart prosthetic valve sound diagnosis apparatus and program - Google Patents

Description

  The present invention relates to an apparatus and a program for diagnosing a state of a heart prosthetic valve, particularly a malfunction, by a sound emitted from the heart prosthetic valve. In the present specification, the malfunction includes a state in which the artificial valve is not functioning normally and a state in which the artificial valve is expected not to function normally.

  Currently, there are over 10,000 cases of heart valve replacement surgery due to infective endocarditis or abnormal myocardial enlargement, and undergoing heart valve replacement surgery annually in Japan. However, it is known that a heart prosthetic valve (hereinafter also simply referred to as a prosthetic valve) can cause malfunction due to blood coagulation, infiltration of living tissue, and the like with long-term use. Therefore, the number of patients undergoing cardiac prosthetic valve replacement that replaces the prosthetic valve is currently about 2500 per year.

Currently, X-ray fluoroscopy has been put to practical use as a method for diagnosing the function of an artificial valve. In addition, a method of applying spectrum analysis (see Non-Patent Documents 1 to 3 below) and a method of applying a Wigner distribution (see Non-Patent Document 4) have been proposed.
Y. Kagawa, N. Sato, S. Nitta, T. Hongo, M. Tanaka, H. Mohri, T. Horiuchi, Real-time sound spectroanalysis for diagnosis of malfunctioning prosthetic valves. Journal of Thoracic and Cardiovascular Surgery 79 (1980) 671-679. A. Gomi, Y. Takeuchi, Y. Okamura, S. Torii, H. Mori, S. Yokoyama, T. Okada, T. Koyama, N. Yamate, The experimental studies to determine the frequency band for the analysis of thrombosed valves Journal of Artificial Organs 21 (1992) 1334-1338. (In Japanese) N. Sato, M. Miura, M. Itoh, M. Ohmi, K. Haneda, H. Mohri, S. Nitta, M. Tanaka. Sound spectral analysis of prosthetic valvular clicks for diagnosis of thrombosed Bjork-Shiley tilting standard disc valve prostheses.Journal of Thoracic and Cardiovascular Surgery 105 (1993) 313-320. J. Hasegawa, K. Kobayashi, Time-frequency domain analysis of the acoustic bio-signal -Successful cases of Wigner distribution applied in medical diagnosis.IEICE Transactions on Fundamentals Electronics, Communications, and Computer Sciences E77-A (1994) 1867-1869 .

  However, spectral analysis cannot take into account the time variation of the spectrum and the S / N ratio is poor. The method of applying the Wigner distribution is not easy, the reliability is not sufficient, the operation is not simple, and the method can be performed frequently. It has not yet been put to practical use because it is not.

  Although X-ray fluoroscopy has been put into practical use, the equipment for that purpose is a large, expensive (hundreds of millions of yen) equipment, and is installed only in limited medical facilities in Japan. Therefore, there is a problem that the inspection requires a large amount of cost and time. In addition, since X-rays are used, there is a drawback of being invasive to a living body.

  As described above, artificial valve dysfunction is difficult to detect at an early stage because there is no simple method for inspecting the valve in daily medical care. Therefore, it is required to be able to diagnose dysfunction early and easily.

  In order to solve the above-described problems, an object of the present invention is to provide a heart prosthetic valve sound diagnostic apparatus and program capable of diagnosing its function non-invasively and quickly from operation sounds generated by an artificial valve. And

  The object of the present invention is achieved by the following means.

That is, the heart prosthetic valve sound diagnosis apparatus according to the present invention is an apparatus that calculates an evaluation value that can be used for diagnosis of a malfunction of a heart prosthetic valve,
A measurement unit for measuring the sound of a heart having an artificial valve;
An analysis unit that performs a short-time Fourier transform on the measured sound,
The analysis unit is
Obtain a spectrogram by squaring the value obtained by the short-time Fourier transform,
The average value of the spectrogram that is within the first period before the maximum peak position in the sound and within the first high frequency range of 1 kHz or more is obtained as HF _pre ,
An average value of the spectrogram in the second period after the maximum peak position and in the second high frequency range of 1 kHz or more is obtained as HF _post ,
A value obtained by dividing the HF _pre by the HF _post is determined as the evaluation value.

  In the above heart prosthetic valve sound diagnosis apparatus, the first period is a period from 10 ms before the maximum peak position to the maximum peak position, and the second period is a period from the maximum peak position to the maximum peak position. It is a period until after 50 ms, and the first and second frequency ranges may be in a range of 5 kHz to 20 kHz.

  The heart valve prosthesis diagnosis device further includes a recording unit that records a reference value, and the prosthetic valve malfunctions depending on whether or not the evaluation value has decreased by a predetermined value or more from the reference value. Can be diagnosed.

  In the above heart prosthetic valve sound diagnostic apparatus, the reference value may be a value determined from a plurality of the evaluation values obtained in a state where the heart prosthetic valve is functioning normally.

  In the above heart valve prosthetic sound diagnosis apparatus, the short-time Fourier transform is performed by setting the sound as x (t) and the window function as w (t).

Thus, STFT (t, ω) can be obtained.

The heart valve prosthesis diagnosis program according to the present invention is a computer-readable program for calculating an evaluation value that can be used for diagnosis of a heart valve malfunction.
On the computer,
A first function for measuring the sound of a heart having an artificial valve;
A second function for performing a short-time Fourier transform on the measured sound;
A third function for obtaining a spectrogram by squaring the value obtained by the short-time Fourier transform;
A fourth function that obtains an average value of the spectrogram that is within a first period before the maximum peak position in the sound and is within a first high frequency range of 1 kHz or more and sets it as HF _pre ;
A fifth function that obtains an average value of the spectrogram within a second period after the maximum peak position and within a second high-frequency range of 1 kHz or more and sets it as HF _post ;
And a sixth function of determining a value obtained by dividing the HF _pre by the HF _post as the evaluation value.

  In the above heart prosthetic valve sound diagnosis program, the first period is a period from 10 ms before the maximum peak position to the maximum peak position, and the second period is a period from the maximum peak position to the maximum peak position. It is a period until after 50 ms, and the first and second frequency ranges may be in a range of 5 kHz to 20 kHz.

  The heart valve prosthesis diagnosis program further realizes a function of recording a reference value in the computer, and determines whether the evaluation value of the prosthetic valve depends on whether or not the evaluation value has decreased by a predetermined value or more from the reference value. Dysfunction can be diagnosed.

  In the above heart prosthetic valve sound diagnosis program, the reference value may be a value determined from a plurality of the evaluation values obtained in a state where the heart prosthetic valve is functioning normally.

  Further, in the above heart prosthetic valve sound diagnosis program, the short-time Fourier transform is such that the sound is x (t) and the window function is w (t).

Thus, STFT (t, ω) can be obtained.

  According to the present invention, it is possible to obtain an evaluation value HFR that can be used for diagnosing a function of a heart prosthetic valve non-invasively, easily and quickly. Therefore, the doctor who is presented with the newly obtained HFR can diagnose the malfunction of the prosthetic valve by comparing the HFR with the past HFR.

  Moreover, if the present invention is realized as a computer program or a portable device, a person having an artificial valve in the heart can easily grasp the state of the artificial valve.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, embodiments of the invention will be described with reference to the accompanying drawings.

  FIG. 1 is a schematic configuration diagram showing a heart prosthetic valve sound diagnostic apparatus (hereinafter also simply referred to as a diagnostic apparatus) according to an embodiment of the present invention. The diagnostic apparatus includes a microphone 2 attached to the stethoscope 1, a measurement unit 3 that collects an output signal of the microphone 2 transmitted via an electric cable (not shown), and a recording unit 4 that records data. , An analysis unit 5 for analyzing the collected data, and a data bus 6 for transmitting data between the measurement unit 3, the recording unit 4, and the analysis unit 5.

When the chest piece 7 of the stethoscope 1 is brought into contact with the chest of a subject having an artificial valve in the heart, the microphone 2 converts the heart sound into an electrical signal and outputs it. Therefore, the microphone 2 can be disposed at a position suitable for efficiently collecting the heart sound of the subject detected by the chest piece 7. In consideration of the size and sensitivity of the microphone 2, for example, the microphone 2 can be disposed at the junction between the chest piece 7 and the tube 8 or in the middle of the tube 8. The arrangement position of the microphone 2 is well known from Japanese Patent Application Laid-Open No. 2005-525521, Japanese Patent Application Laid-Open No. 2008-206593, and the like, and thus the description thereof is omitted.

  The heart valve replaced with the artificial valve is mainly a mitral valve and an aortic valve, and their positions are different. Therefore, when measuring heart sounds in the present invention, it is desirable that the chest piece 7 is brought into contact with an appropriate position of the subject's chest in consideration of the replaced artificial valve. This can be easily done by a physician.

  The measurement unit 3 samples the analog signal from the microphone 2 at a predetermined time interval and collects it as digital data. The signal from the microphone 2 may be collected after passing through an equalizer or an amplifier.

  For example, a computer can be used as the recording unit 4, the analysis unit 5, and the internal bus 6, and in this case, the analysis unit 5 includes an arithmetic processing device (CPU), a temporary storage device (RAM), and the like. Furthermore, if a computer expansion board having the function of the measurement unit 3 is mounted on the internal bus 6, they can be configured integrally.

  For example, the analysis unit 5 reads data to be analyzed from the recording unit 4 and executes an analysis process to be described later. Using this analysis result, for example, a doctor can determine whether or not the heart prosthetic valve is functioning normally.

  The operation of the heart prosthetic valve sound diagnosis apparatus will be specifically described based on the flowchart shown in FIG. In the following description, the analysis unit 5 uses the internal temporary storage device as a storage area for data read from the recording unit 4, a calculation work area, a storage area for calculation results, and the like. To do.

  In step S <b> 1, the analysis unit 5 receives an input of information (hereinafter referred to as “subject code”) that identifies a subject to be measured. If the code has not been input in the past, it is newly recorded in the recording unit 4. For the input means, a computer keyboard, a mouse, or the like can be used.

  In step S2, the measuring unit 3 collects the signal from the microphone 2 as described above. An example of the collected data is shown in FIG. The vertical axis and horizontal axis represent amplitude (voltage) and time, respectively. The middle and lower graphs in FIG. 3 show the artificial valve sound (heart sound from the heart replaced with the artificial valve), and the upper part shows the normal heart sound (heart sound of a person who has not undergone heart valve replacement). FIG. 3 shows a range of −20 ms to +40 ms with the maximum peak position of the heart sound as a reference (time 0 ms).

In step S3, the analysis unit 5 performs spectrogram analysis on the data collected in step S2. Specifically, the analysis unit 5 converts the read data by STFT (Short-Time Fourier Transformation) represented by the following equation, and squares the converted value to obtain a spectrogram.

Here, w (t) is a window function for processing data in a predetermined time range from the measurement data. For the window function, a commonly used known Hamming window, Gaussian window, Blackman window, or the like can be used. x (t) is data before conversion, and STFT (t, ω) (also referred to as X (t, ω)) is data after conversion. X (t, ω) is also expressed as STFT {x (t)} using the data x (t) before conversion. STFT {x (t)}, that is, X (t, ω) is a Fourier transform coefficient of x (τ) w (τ−t), and represents the phase and amplitude of the signal in time and frequency space.

The spectrogram is obtained by | X (t, ω) | ² . FIG. 4 shows an example of a spectrogram. The vertical axis represents frequency f (kHz) (f = ω / (2π)), and the horizontal axis represents time t (ms). Each stage diagram shows the result obtained by converting the data of the corresponding stage in FIG. It can be seen from the upper data in FIG. 4 that the normal heart sound includes only frequency components of less than 1 kHz. On the other hand, it can be seen from the middle and lower data that the artificial valve sound also includes a high frequency component of 1 kHz to 20 kHz. This result is consistent with the findings of Non-Patent Document 2.

In step S4, the analysis unit 5 analyzes the spectrogram | X (
From t, ω) | ² , an evaluation value HFR (High Frequency Rate) is obtained by the following equation.
HFR = 10 log [HF _pre / HF _post ] [dB]
Here, HF _pre is an average spectrogram in a frequency range of 5 kHz to 20 kHz in a period (−10 ms to 0 ms) from 10 ms before the maximum peak position of the heart sound waveform (see FIG. 3) to the maximum peak position. HF _post is an average spectrogram in a frequency range of 5 kHz to 20 kHz in a period (0 ms to 50 ms) from the maximum peak position of the heart sound waveform to 50 ms thereafter. HF _pre and HF _post are obtained as decibel values. The calculated HFR is recorded in the recording unit 4 in association with the subject code. At this time, it is desirable to add and record information on the date on which the heart sound data was collected in step S2.

  In step S5, it is determined whether or not the HFR corresponding to the subject code input in step S1 is recorded in the recording unit 4. If it has been recorded, the process proceeds to step S6. If it has not been recorded, the process ends.

In step S6, the past data (HFR obtained when the artificial valve is functioning normally) recorded in the recording unit 4 in correspondence with the subject code input in step S1 is read, and in step S4. Compare with the calculated HFR and present the result. For example, the difference ΔHFR = HFR _av −HF between the average value HFR _av of the past HFR and the HFR ₀ calculated this time
Present R ₀ . For example, the difference ΔHFR is displayed on a display device such as a liquid crystal or a CRT.

  As described above, HFR represents a high-frequency component generated by the artificial valve. Therefore, when a thrombus adheres to the artificial valve, the generated high-frequency component decreases. Therefore, if the heart sound is regularly measured for the same subject, and the HFR is calculated and recorded, the doctor can determine whether the newly calculated HFR has decreased from the past HFR. Valve dysfunction can be diagnosed. For example, if ΔHFR ≦ a, it can be determined that the artificial valve is functioning normally, and if the HFR is out of this range, it can be determined that the artificial valve is malfunctioning. The reference value a can be set from the degree of fluctuation of HFR obtained periodically in a state where the artificial valve is functioning normally. For example, a = 3 dB.

Although the case where the difference ΔHFR between the average value of the past HFR and the newly calculated HFR is presented for the same subject has been described above, the present invention is not limited to this. A representative value of a plurality of past HFRs may be used. For example, an intermediate value may be used instead of the average value. Further, past HFRs and newly calculated HFRs may be combined and presented, for example, as a time-series graph. In this case, the doctor can diagnose the malfunction of the prosthetic valve from the change tendency of HFR.

  Even if the prosthetic valve is functioning normally, it is considered that the HFR changes depending on the individual. Therefore, it is desirable to set an evaluation standard for each individual (for example, the above-described reference value a is set for each individual). . However, when a malfunction occurs in the prosthetic valve, an urgent re-replacement operation is required, and the above-mentioned measurement cannot be performed in many cases. Therefore, it is not easy to set an evaluation standard for each individual. Therefore, as described above, if the difference ΔHFR is used, the reference value a obtained from the data of another subject or the reference value a obtained from the data of a plurality of different subjects can be used regardless of the subject. In addition, it is possible to diagnose the malfunction of the artificial valve.

In the above description, the value obtained by dividing HF _post without using HF _{pre as it is} is used as the evaluation value HFR in order to remove the influence of noise and normalize (normalize). Since the noise is considered to change depending on the measurement state, the diagnosis reliability is determined by dividing the HF _pre by the HF _{post in the} region (specified in the frequency band and the time band) in which only the noise is included without generating a high frequency component. Is important to.

Moreover, HF _pre is not limited to the value calculated in the region where the time band is −10 ms to 0 ms and the frequency band is 5 kHz to 20 kHz as described above. The time zone is near the maximum peak position of the heart sound waveform and can be a predetermined time zone before the maximum peak position. Further, the frequency band can be a predetermined high frequency band including a frequency that is not included in the normal heart sound, for example, 1 kHz or more and 20 kHz or less. Similarly, HF _post is not limited to a value calculated in a region where the time band is 0 ms to 50 ms and the frequency band is 5 kHz to 20 kHz. The time zone can be a predetermined time zone that does not include high frequency components. The frequency band is the same as that of HF _post .

Further, the evaluation value HFR may not be a decibel value. For example, HF _pre / HF _post
It is good also as HFR as it is.

  Further, the spectrogram analysis is not limited to the above-described STFT, and any method that can obtain a spectrogram in a predetermined period before and after the maximum peak of the heart sound may be used.

  In addition, the diagnosis apparatus is installed in a medical institution such as a hospital and is not limited to a diagnosis by a doctor, but a person having a heart prosthetic valve can also make a diagnosis. For example, if the above function is realized by a computer program and the computer program is installed in a personally owned computer, the individual can regularly measure and record the obtained results (such as HFR), You can judge the state of the artificial valve yourself. It can also be realized as a portable dedicated device. In those cases, the stethoscope does not need to be a normal stethoscope, but only needs to include at least a chest piece and a microphone.

  Alternatively, only the heart sound may be measured by an individual, the measurement data may be transmitted to a medical institution via a communication means such as the Internet, and the analysis and HFR calculation may be performed by the medical institution.

  Examples are shown below to further clarify the features of the present invention.

As described above, heart sound data was measured for 18 subjects, and HFR was obtained. Out of 18 people (31 to 90 years old), 16 people (5 men and 11 women, represented by Sub1 to 8 and Sub10 to 17) have mitral or aortic valves replaced with prosthetic valves. A person (male, represented by Sub9 and 18) does not have a heart valve. The heart sound was measured at a position where the heart prosthetic valve sound was maximized by directly placing a chestpiece of a stethoscope on the chest of a subject in a supine state without using clothes. For the stethoscope, a Littmann Master Classic II Stethoscope (manufactured by Sumitomo 3M) was used, and a small microphone (AT805F, manufactured by Audio Technica) was placed in the tube at the junction with the chestpiece. . The signal was collected from the microphone, that is, recorded, using an A / D converter (UA-1000, manufactured by Ediroll), data was collected at a sampling frequency of 44.1 kHz and 16 bits, and recorded on a hard disk of a computer. . The recording time was about 10 to 20 seconds for each subject, and measurements were performed 10 times or more. The results are shown in FIGS.

  FIG. 5 is a graph showing an example of a measured heart sound waveform. (A) is a normal heart sound, and (b) and (c) are artificial valve sounds. The artificial valve sound includes two main sounds. They are the first sound generated by blocking the backflow of blood by closing the mitral and tricuspid valves early in the ventricular contraction, and the aortic and pulmonary valves at the end of the ventricular contraction. This is a second sound generated by blocking the backflow of blood by closing.

  FIG. 6 is a diagram showing a spectrogram obtained by processing the data of FIG. 5 as described above (see step S3).

FIG. 7 is a graph showing HF _pre and HF _post determined as described above from the spectrograms obtained for all subjects (Sub 1 to 18) (see step S4). A black circle represents HF _pre, and a white circle represents HF _post . Each is represented by an average value obtained from 10 times of measurement data and an error bar. These differ statistically by a significant difference of 1%. There are individual differences in the HF _pre and HF _post of the artificial valve sound, which is due to the difference in the type of heart prosthetic valve, fat thickness, blood pressure, etc. depending on the subject.

From FIG. 7, it can be seen that HF _pre and HF _post have almost the same value for the subject (Sub 9, 18) that does not have a heart prosthetic valve. On the other hand, for subjects (Sub 1 to 8, Sub 10 to 17) having a heart prosthetic valve, HF _pre and HF _post are different values, and HF _pre is larger than HF _post . From these facts, it was confirmed that a high-frequency component due to the artificial valve was included in 10 ms (−10 to 0 ms) immediately before the maximum peak of the heart sound.

FIG. 8 is a graph showing HFR obtained as described above from HF _pre and HF _post obtained for all subjects (Sub 1 to 18) (see step S4). From FIG. 8, it can be seen that HFR = 0 for subjects (Sub 9, 18) that do not have a heart prosthetic valve. On the other hand, it can be seen that HFR is a positive value different from 0 for subjects (Sub 1 to 8, Sub 10 to 17) having a heart prosthetic valve. Therefore, it can be said that the high-frequency component contained in the heart valve sound can be quantified.

  As described above, according to the present invention, since the high-frequency component contained in the artificial valve sound can be quantified by the HFR, it can be seen that the HFR is effective for diagnosing the malfunction of the heart prosthetic valve. For example, HFR is regularly observed and recorded, and when the newly obtained HFR greatly changes from the past HFR, it can be diagnosed that the heart valve is malfunctioning.

1 is a schematic configuration diagram showing a heart valve prosthesis diagnosis device according to an embodiment of the present invention. It is a flowchart explaining operation | movement of this diagnostic apparatus. It is a graph which shows an example of the signal extract | collected with the microphone. It is a figure which shows a spectrogram. It is a graph which shows an example of the heart sound obtained in the Example. It is a figure which shows an example of the spectrogram obtained in the Example. It is a graph which shows HF _pre and HF _post obtained about each test subject of the Example. It is a graph which shows HFR obtained about each subject of an example.

Explanation of symbols

1 Stethoscope 2 Microphone 3 Measuring unit 4 Recording unit 5 Analysis unit 6 Internal bus 7 Chest piece 8 Tube and ear canal

Claims (10)

An apparatus for calculating an evaluation value that can be used for diagnosis of malfunction of a cardiac prosthetic valve,
A measurement unit for measuring the sound of a heart having an artificial valve;
An analysis unit that performs a short-time Fourier transform on the measured sound,
The analysis unit is
Obtain a spectrogram by squaring the value obtained by the short-time Fourier transform,
The average value of the spectrogram that is within the first period before the maximum peak position in the sound and within the first high frequency range of 1 kHz or more is obtained as HF _pre ,
An average value of the spectrogram in the second period after the maximum peak position and in the second high frequency range of 1 kHz or more is obtained as HF _post ,
A heart prosthetic valve sound diagnosis apparatus characterized in that a value obtained by dividing the HF _pre by the HF _post is determined as the evaluation value. The first period is a period from 10 ms before the maximum peak position to the maximum peak position;
The second period is a period from the maximum peak position to 50 ms after the maximum peak position;
The cardiac prosthetic valve sound diagnosis apparatus according to claim 1, wherein the first and second frequency ranges are in a range of 5 kHz to 20 kHz. It further comprises a recording unit that records the reference value,
3. The heart valve prosthesis diagnosis device according to claim 1, wherein a malfunction of the artificial valve is diagnosed based on whether or not the evaluation value has decreased by a predetermined value or more from the reference value.   The heart valve prosthesis diagnosis device according to claim 3, wherein the reference value is a value determined from a plurality of the evaluation values obtained in a state where the heart prosthetic valve is functioning normally. The short-time Fourier transform has the sound as x (t) and the window function as w (t).
The heart valve prosthesis diagnosis device according to any one of claims 1 to 4, wherein STFT (t, ω) is obtained by the following process. A computer-readable program for calculating an evaluation value that can be used for diagnosis of malfunction of a cardiac prosthetic valve,
On the computer,
A first function for measuring the sound of a heart having an artificial valve;
A second function for performing a short-time Fourier transform on the measured sound;
A third function for obtaining a spectrogram by squaring the value obtained by the short-time Fourier transform;
A fourth function that obtains an average value of the spectrogram that is within a first period before the maximum peak position in the sound and is within a first high frequency range of 1 kHz or more and sets it as HF _pre ;
A fifth function that obtains an average value of the spectrogram within a second period after the maximum peak position and within a second high-frequency range of 1 kHz or more and sets it as HF _post ;
A heart prosthetic valve sound diagnosis program for realizing a sixth function of determining a value obtained by dividing the HF _pre by the HF _post as the evaluation value. The first period is a period from 10 ms before the maximum peak position to the maximum peak position;
The second period is a period from the maximum peak position to 50 ms after the maximum peak position;
The heart valve prosthesis diagnosis program according to claim 6, wherein the first and second frequency ranges are in a range of 5 kHz to 20 kHz. Further realizing the function of recording a reference value in the computer,
The heart valve prosthesis diagnosis program according to claim 6 or 7, wherein a malfunction of the prosthetic valve is diagnosed based on whether or not the evaluation value has decreased from the reference value by a predetermined value or more.   The heart valve prosthesis diagnosis program according to claim 8, wherein the reference value is a value determined from a plurality of the evaluation values obtained in a state where the heart prosthetic valve is functioning normally. The short-time Fourier transform has the sound as x (t) and the window function as w (t).
The heart prosthetic valve sound diagnosis program according to any one of claims 6 to 9, wherein STFT (t, ω) is obtained by the following process.