Electronic stethoscope signal processing device, electronic stethoscope system, electronic stethoscope signal processing program, and electronic stethoscope signal processing method

Abstract

To reduce not only stationary noise (low-frequency environmental sound or noise etc.), but also burst noise (a high frequency cry or speaking etc.) so as to extract a body sound (low-frequency heart beat sound or breath sound etc.).SOLUTION: While an improved adaptive filter part 4 and an improved nonlinear filter part 5 are used to reduce burst noise (a high frequency cry or speaking etc.), body sound (low-frequency heat beat sound or breath sound etc.) is maintained. Here, the degree of reduction of the burst noise (the high frequency cry or speaking etc.) and the degree of maintenance of the body sound (the low-frequency heat beat sound or breath sound, etc.) are quantitatively measured.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an electronic stethoscope that reduces external sounds and extracts internal sounds. [Background technology]

[0002] An electronic stethoscope that reduces external sounds and extracts internal sounds is disclosed in Non-Patent Document 1 and elsewhere. Here, an external sound collector of the electronic stethoscope collects external sounds and outputs an external sound signal. On the other hand, an internal sound collector of the electronic stethoscope collects internal sounds and outputs an internal sound signal. However, the internal sound signal also includes components of external sounds that leak from outside the electronic stethoscope to the internal sound collector.

[0003] The adaptive filter section of the signal processing device has adaptive filter coefficients that simulate the characteristics of extracorporeal sound leaking from the outside of the electronic stethoscope to the internal sound collector, and inputs the extracorporeal sound signal obtained from the internal sound collector and outputs an adaptive filter signal. Furthermore, the residual signal calculation section of the signal processing device subtracts the adaptive filter signal from the internal sound signal obtained from the internal sound collector and calculates a residual signal in which the leaking components of the extracorporeal sound signal have been reduced from the internal sound signal. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Samir B. Patel, Thomas F. Callahan, Matthew G. Callahan, James T. Jones, George P. Graber, Kirk S. Foster, Kenneth Glifort, and George R. Wodicka, “An adaptive noise reduction stethoscope for auscultation in high noise environments”, The Journal of the Acoustical Society of America, vol. 103, no. 5, pp. 2483-2491, 1998. Summary of the Invention [Problem to be solved by the invention]

[0005] Here, the adaptive filter unit and residual signal calculation unit of the signal processing device can reduce stationary noise (low-frequency environmental sounds or noises, etc.) and extract internal body sounds (low-frequency heartbeat sounds or breathing sounds, etc.). However, the adaptive filter unit and residual signal calculation unit of the signal processing device cannot sufficiently reduce sudden noise (high-frequency crying or talking, etc.) and extract internal body sounds (low-frequency heartbeat sounds or breathing sounds, etc.). The reasons for this will be described later using Figures 2 to 4.

[0006] Therefore, in order to solve the above-mentioned problems, the present disclosure aims to extract internal body sounds (low-frequency heartbeat sounds, breathing sounds, etc.) by reducing not only steady noises (low-frequency environmental sounds or noises, etc.), but also sudden noises (high-frequency crying or talking, etc.). [Means for solving the problem]

[0007] To solve the above problem, in an adaptive filter, a provisional filter unit has provisional filter coefficients set equal to updated values ​​of the adaptive filter coefficients, and inputs an extracorporeal sound signal to output a provisional filter signal.The correction gain calculation unit calculates a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients to be a relatively larger value for a band in which the correlation in the frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the provisional filter signal is higher (high-frequency crying or talking, etc.).On the other hand, the correction gain calculation unit calculates a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients to be a relatively smaller value for a band in which the correlation in the frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the provisional filter signal is lower (low-frequency heartbeat or breathing sounds, etc.).

[0008] Specifically, the present disclosure provides an adaptive filter unit having adaptive filter coefficients that simulate the characteristics of extracorporeal sound leaking from the outside of an electronic stethoscope to an intracorporeal sound collector, and that inputs an extracorporeal sound signal obtained from the extracorporeal sound collector and outputs an adaptive filter signal; a residual signal calculation unit that subtracts the adaptive filter signal from the intracorporeal sound signal obtained from the intracorporeal sound collector and calculates a residual signal from the intracorporeal sound signal in which the leak component of the extracorporeal sound signal has been reduced; a filter coefficient update unit that adds an updated value of the adaptive filter coefficient to a current value of the adaptive filter coefficient so as to reduce the correlation in the time axis direction between the adaptive filter signal and the residual signal; and a provisional filter unit having provisional filter coefficients set equal to the updated value of the adaptive filter coefficient, and that inputs the extracorporeal sound signal and outputs a provisional filter signal. a correction gain calculation unit that calculates a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients to a relatively larger value for a band in which the correlation in the frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the provisional filter signal is higher, and calculates a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients to a relatively smaller value for a band in which the correlation in the frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the provisional filter signal is lower; and a filter coefficient correction unit that calculates a correction value for the updated values ​​of the adaptive filter coefficients based on the correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients.

[0009] This configuration uses an adaptive filter to reduce sudden noises (high-frequency sounds such as crying or talking) while preserving internal body sounds (low-frequency sounds such as heartbeat or breathing sounds).

[0010] The present disclosure also provides an electronic stethoscope signal processing device, characterized in that the correction gain calculation unit calculates, for each band, a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients, based on the correlation in the frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the provisional filter signal obtained by multiplying, for each band, a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients, so as to minimize a shape error between the two frequency spectra for each band.

[0011] According to this configuration, it is possible to quantitatively calculate the degree to which sudden noise (high-frequency crying or talking, etc.) is reduced and the degree to which internal body sounds (low-frequency heartbeat sounds or breathing sounds, etc.) are maintained.

[0012] The present disclosure also provides an electronic stethoscope signal processing device, further comprising: a filter reduction amount calculation unit that calculates the amount of reduction of the frequency spectrum of the residual signal relative to the frequency spectrum of the internal body sound signal and calculates the amount of reduction by the adaptive filter unit; and a residual signal suppression unit that suppresses the frequency spectrum of the residual signal to a greater extent for frequencies with a greater amount of reduction by the adaptive filter unit, while maintaining or suppressing the frequency spectrum of the residual signal to a lesser extent for frequencies with a smaller amount of reduction by the adaptive filter unit.

[0013] This configuration uses a nonlinear filter to reduce sudden noises (such as high-frequency crying or talking) while preserving internal body sounds (such as low-frequency heartbeat or breathing sounds).

[0014] To solve the above problem, in the nonlinear filter, the filter reduction amount calculation unit calculates the amount of reduction of the frequency spectrum of the residual signal relative to the frequency spectrum of the body sound signal, and calculates the amount of reduction by the adaptive filter unit.The residual signal suppression unit then suppresses the frequency spectrum of the residual signal more to a greater extent for frequencies where the amount of reduction by the adaptive filter unit is greater (high-frequency crying or talking, etc.).On the other hand, the residual signal suppression unit preserves the frequency spectrum of the residual signal more to a lesser extent for frequencies where the amount of reduction by the adaptive filter unit is smaller (low-frequency heartbeat sounds or breathing sounds, etc.).

[0015] Specifically, the present disclosure provides an electronic stethoscope signal processing device comprising: an adaptive filter unit having adaptive filter coefficients that simulate the leakage characteristics of extracorporeal sound from outside the electronic stethoscope to an intracorporeal sound collector, and that inputs an extracorporeal sound signal obtained from the extracorporeal sound collector and outputs an adaptive filter signal; a residual signal calculation unit that subtracts the adaptive filter signal from the intracorporeal sound signal obtained from the intracorporeal sound collector and calculates a residual signal in which the leakage components of the extracorporeal sound signal have been reduced from the intracorporeal sound signal; a filter coefficient update unit that adds an updated value of the adaptive filter coefficient to a current value of the adaptive filter coefficient so as to reduce the correlation in the time axis direction between the adaptive filter signal and the residual signal; a filter reduction amount calculation unit that calculates the amount of reduction of the frequency spectrum of the residual signal with respect to the frequency spectrum of the intracorporeal sound signal and calculates the amount of reduction by the adaptive filter unit; and a residual signal suppression unit that suppresses the frequency spectrum of the residual signal more strongly for frequencies where the amount of reduction by the adaptive filter unit is greater, while maintaining or reducing the frequency spectrum of the residual signal more strongly for frequencies where the amount of reduction by the adaptive filter unit is smaller.

[0016] This configuration uses a nonlinear filter to reduce sudden noises (such as high-frequency crying or talking) while preserving internal body sounds (such as low-frequency heartbeat or breathing sounds).

[0017] The present disclosure also provides an electronic stethoscope signal processing device, further comprising a residual signal minimum smoothing unit that calculates a minimum smoothed spectrum of the residual signal so as to maintain a minimum value of the frequency spectrum of the residual signal, wherein the residual signal suppression unit is configured to suppress the frequency spectrum of the residual signal more to a greater extent toward the minimum smoothed spectrum of the residual signal for frequencies at which the amount of reduction by the adaptive filter unit is greater, while maintaining the frequency spectrum of the residual signal or suppressing it more to a lesser extent toward the minimum smoothed spectrum of the residual signal for frequencies at which the amount of reduction by the adaptive filter unit is smaller.

[0018] According to this configuration, it is possible to quantitatively calculate the degree to which sudden noise (high-frequency crying or talking, etc.) is reduced and the degree to which internal body sounds (low-frequency heartbeat sounds or breathing sounds, etc.) are maintained.

[0019] The present disclosure also provides an electronic stethoscope system comprising the electronic stethoscope signal processing device described above, an extracorporeal sound collector that collects extracorporeal sounds and outputs the extracorporeal sound signals, and an internal sound collector that collects internal body sounds and outputs the internal body sound signals.

[0020] According to this configuration, by reducing sudden noises (high-frequency crying or talking, etc.), it is possible to extract internal body sounds (low-frequency heartbeat sounds, breathing sounds, etc.).

[0021] The present disclosure also provides an electronic stethoscope signal processing program for causing a computer to execute each processing step performed by each component of the electronic stethoscope signal processing device described above.

[0022] According to this configuration, by reducing sudden noises (high-frequency crying or talking, etc.), it is possible to extract internal body sounds (low-frequency heartbeat sounds, breathing sounds, etc.).

[0023] The present disclosure also provides an electronic stethoscope signal processing method comprising the processing steps performed by the components of the electronic stethoscope signal processing device described above.

[0024] According to this configuration, by reducing sudden noises (high-frequency crying or talking, etc.), it is possible to extract internal body sounds (low-frequency heartbeat sounds, breathing sounds, etc.). [Effects of the Invention]

[0025] In this way, the present disclosure not only reduces steady noise (such as low-frequency environmental sounds or noises), but also reduces sudden noise (such as high-frequency crying or talking), thereby making it possible to extract internal body sounds (such as low-frequency heartbeat sounds or breathing sounds). [Brief explanation of the drawings]

[0026] [Figure 1] FIG. 1 illustrates components of the electronic stethoscope system of the present disclosure. [Figure 2] FIG. 1 illustrates a problem to be solved by the electronic stethoscope system of the present disclosure. [Figure 3] FIG. 1 illustrates a problem to be solved by the electronic stethoscope system of the present disclosure. [Figure 4] FIG. 1 illustrates a problem to be solved by the electronic stethoscope system of the present disclosure. [Figure 5] FIG. 2 is a diagram illustrating components of an improved adaptive filter section of the present disclosure. [Figure 6] FIG. 10 is a diagram illustrating a processing procedure of an improved adaptive filter unit according to the present disclosure. [Figure 7] FIG. 2 is a diagram illustrating the processing concept of an improved adaptive filter unit of the present disclosure. [Figure 8] FIG. 2 illustrates components of the improved nonlinear filter section of the present disclosure. [Figure 9] FIG. 10 is a diagram illustrating a processing procedure of an improved nonlinear filter unit according to the present disclosure. [Figure 10] FIG. 2 is a diagram illustrating the processing concept of an improved nonlinear filter unit of the present disclosure. [Figure 11] 10A and 10B are diagrams illustrating the processing results of the electronic stethoscope signal processing device of the present disclosure. [Figure 12] 1A and 1B are diagrams showing the processing results of a conventional and presently disclosed electronic stethoscope signal processing device. DETAILED DESCRIPTION OF THE INVENTION

[0027]

[0023] The following embodiments of the present disclosure will be described with reference to the accompanying drawings. The embodiments described below are examples of implementation of the present disclosure, and the present disclosure is not limited to the following embodiments.

[0028] (Problems to be solved by the electronic stethoscope system of the present disclosure) The components of the electronic stethoscope system of the present disclosure are shown in Figure 1. The electronic stethoscope system S comprises an electronic stethoscope E and an electronic stethoscope signal processing device P. The electronic stethoscope E comprises an external sound collector 1 and an internal sound collector 2. The electronic stethoscope signal processing device P comprises an external and internal sound signal input unit 3, an improved adaptive filter unit 4, an improved nonlinear filter unit 5, and an internal sound signal output unit 6, and can be realized by installing the electronic stethoscope signal processing programs shown in Figures 6 and 9 on a computer.

[0029] The electronic stethoscope E is a sensor device for collecting internal body sounds via the body surface B and acquiring an internal body sound signal. The external sound collector 1 collects external body sounds and outputs an external body sound signal. The internal body sound collector 2 collects internal body sounds and outputs an internal body sound signal. Examples of external body sounds include low-frequency environmental sounds or noises and high-frequency crying or talking. Examples of internal body sounds include low-frequency heartbeat sounds or breathing sounds. However, the internal body sound signal also includes components of external body sounds that are transmitted from outside the electronic stethoscope E to the internal body sound collector 2.

[0030] 2 to 4 show the problem to be solved by the electronic stethoscope system of the present disclosure. In FIG. 2, the relative transfer gain of extracorporeal sound leaking from the outside of electronic stethoscope E to internal sound collector 2 is greater in the lower frequency range and smaller in the higher frequency range (the difference is, for example, about 40 dB). Therefore, the reduction effect of the adaptive filter of the prior art is greater in the lower frequency range and smaller in the higher frequency range. On the other hand, equal loudness curves (contour lines at which the human ear perceives the same loudness of sound) have not only a high sensitivity band on the order of several tens of Hz to 100 Hz, but also a high sensitivity band on the order of several hundreds to several thousand Hz. Therefore, even if the relative transfer gain of extracorporeal sound leaking from the outside of electronic stethoscope E to internal sound collector 2 is small in the high frequency range, if the reduction effect of the adaptive filter of the prior art is small in the high frequency range, the high-frequency extracorporeal sound will sound loud to the human ear and interfere with the hearing of low-frequency internal sounds.

[0031] Figure 3 shows a schematic diagram of the output signal of a conventional adaptive filter. In the low frequency range where the transfer gain of extracorporeal sound leakage is high, the adaptive filter coefficients are easily updated, and the simulation accuracy of the extracorporeal sound leakage components in the internal sound signal is high. In the high frequency range where the transfer gain of extracorporeal sound leakage is low, the adaptive filter coefficients are not easily updated, and the simulation accuracy of the extracorporeal sound leakage components in the internal sound signal is low. Therefore, as shown in Figures 5 to 7, processing by the improved adaptive filter unit 4 is performed.

[0032] Figure 4 shows a schematic diagram of the suppression effect of a conventional nonlinear filter. The upper part of Figure 4 shows the suppression effect of a nonlinear filter that deepens the spectral valleys. The suppression effect of the nonlinear filter from the residual signal to the suppressed signal results in the desired noise removal in the low frequency range, but unintended noise enhancement in the high frequency range. The lower part of Figure 4 shows the suppression effect of a nonlinear filter that flattens the spectral peaks. The suppression effect of the nonlinear filter from the residual signal to the suppressed signal results in the desired noise removal in the high frequency range, but unintended signal removal in the low frequency range. Therefore, the processing of the improved nonlinear filter unit 5 is performed as shown in Figures 8 to 10.

[0033] (Processing Concept of the Improved Adaptive Filter Unit of the Present Disclosure) The components of the improved adaptive filter unit of the present disclosure are shown in Figure 5. The processing procedure of the improved adaptive filter unit of the present disclosure is shown in Figure 6. The processing concept of the improved adaptive filter unit of the present disclosure is shown in Figure 7. The external and internal sound signal input unit 3 includes an external sound signal AD conversion unit 31, an internal sound signal AD conversion unit 32, and an internal sound signal delay unit 33. The improved adaptive filter unit 4 includes an adaptive filter unit 41, a residual signal calculation unit 42, a filter coefficient update unit 43, a temporary filter unit 44, a correction gain calculation unit 45, and a filter coefficient correction unit 46. The external and internal sound signal input unit 3 and the improved adaptive filter unit 4 can be realized by installing the improved adaptive filter program shown in Figure 6 on a computer.

[0034] The external sound signal AD conversion unit 31 outputs the external sound signal x(n) after AD conversion (step S1). The internal sound signal AD conversion unit 32 outputs the internal sound signal y(n) after AD conversion (step S2). The internal sound signal delay unit 33 delays the internal sound signal y(n) by a certain delay D. D (n)=y(nD) is output (step S3).

[0035] The adaptive filter unit 41 has an adaptive filter coefficient w(m) that simulates feedback from the external sound collector 1 to the internal sound collector 2, receives the external sound signal x(n) as input, and outputs an adaptive filter signal d(n)=Σw(m)x(nm) (the sum is calculated for m=0 to M) (step S4), where M is the order of the adaptive filter unit 41. The adaptive filter unit 41 may also perform processing in the frequency domain that is equivalent to that in the time domain.

[0036] The residual signal calculation unit 42 calculates the internal body sound signal y D (n), the adaptive filter signal d(n) (a simulated signal of the feedback component of the external sound signal x(n)) is subtracted from the internal sound signal y(n). D (n) is the residual signal r(n)=y D (n)-d(n) is calculated (step S5).

[0037] The filter coefficient update unit 43 adds the updated value Δw(m) of the adaptive filter coefficient to the current value w(m) of the adaptive filter coefficient so as to reduce the correlation in the time axis direction between the adaptive filter signal d(n) and the residual signal r(n) (step S6, Equation 1). Here, μ and λ are fixed or time-varying control parameters that determine the magnitude of the updated value Δw(m) of the adaptive filter coefficient.

number

[0038] The provisional filter unit 44 has a provisional filter coefficient Δw(m) set equal to the updated value Δw(m) of the adaptive filter coefficient, inputs the extracorporeal sound signal x(n), and outputs a provisional filter signal e(n)=ΣΔw(m)x(nm) (the sum is calculated for m=0 to M) (step S7), where M is the order of the provisional filter unit 44. The provisional filter unit 44 may also perform processing in the frequency domain equivalent to that in the time domain.

[0039] The correction gain calculation unit 45 calculates a relatively larger value for the correction gain g(b) for the frequency spectrum ΔW(k) of the updated value Δw(m) of the adaptive filter coefficient in a band b where the correlation in the frequency axis direction between the frequency spectrum R(k) of the residual signal r(n) and the frequency spectrum E(k) of the tentative filter signal e(n) is higher (steps S8 to S10, the high bands in the upper and middle parts of FIG. 7).

[0040] On the other hand, the correction gain calculation unit 45 calculates a relatively smaller value for the correction gain g(b) for the frequency spectrum ΔW(k) of the updated value Δw(m) of the adaptive filter coefficient in a band b where the correlation in the frequency axis direction between the frequency spectrum R(k) of the residual signal r(n) and the frequency spectrum E(k) of the tentative filter signal e(n) is lower (steps S8 to S10, the lower bands in the upper and middle rows of FIG. 7).

[0041] Specifically, based on the correlation in the frequency axis direction between the frequency spectrum R(k) of the residual signal r(n) and the frequency spectrum E(k) of the provisional filter signal e(n) obtained by multiplying, for each band b, the correction gain g(b) for the frequency spectrum ΔW(k) of the updated value Δw(m) of the adaptive filter coefficient, the correction gain calculation unit 45 calculates, for each band b, the correction gain g(b) for the frequency spectrum ΔW(k) of the updated value Δw(m) of the adaptive filter coefficient so as to minimize the shape error between the two frequency spectra R(k) and E(k) for each band b (steps S8 to S10, upper and middle parts of FIG. 7: correction gain g(b) for the b-th band among the divided bands).

[0042] First, the amplitude or power spectrum R(k) of the short-time residual signal r(n) is calculated, and the spectra R(k) within the b-th band are collectively vectorized to R(b) (step S8). Next, the amplitude or power spectrum E(k) of the short-time provisional filter signal e(n) is calculated, and the spectra E(k) within the b-th band are collectively vectorized to E(b) (step S9). Next, the correction gain g(b) for the frequency spectrum ΔW(k) of the updated value Δw(m) of the adaptive filter coefficient is calculated (step S10, Equation 2). Here, < > is an inner product.

number

[0043] The filter coefficient correction unit 46 calculates a correction value Δw(m) for the adaptive filter coefficient update value Δw(m) based on a correction gain g(b) for the frequency spectrum ΔW(k) of the adaptive filter coefficient update value Δw(m). new (m) (steps S11 to S14, multiple bands b: high frequency and low frequency in the upper and middle rows of FIG. 7). The filter coefficient update unit 43 calculates the correction value Δw(m) for the updated value Δw(m) of the adaptive filter coefficient without applying the updated value Δw(m) of the adaptive filter coefficient (step S6). new (m) is applied (step S14).

[0044] First, the frequency spectrum ΔW(k) of the updated value Δw(m) of the adaptive filter coefficient is calculated (step S11). Next, the frequency spectrum ΔW(k) of the updated value Δw(m) of the adaptive filter coefficient is calculated as ΔW(k) in the b-th band. new (k)=g(b)ΔW(k) (step S12). Next, the adaptive filter coefficient correction value Δw new Frequency spectrum of (m) ΔW new (k) to Δw new (m) in the time domain (step S13). Next, the updated value Δw(m) of the adaptive filter coefficient is expressed as Δw new (m) (step S14).

[0045] In the high frequency range in the lower part of Figure 7, the frequency spectrum R(k) of the residual signal r(n) contains many external sound loop components before correction with the adaptive filter coefficient update value Δw(m), and after correction with the adaptive filter coefficient update value Δw(m), the external sound loop components are reduced, but not necessarily completely reduced (first row of Figure 11). In the low frequency range in the lower part of Figure 7, the frequency spectrum R(k) of the residual signal r(n) contains mainly internal sound components before correction with the adaptive filter coefficient update value Δw(m), and the internal sound components are almost maintained before and after correction with the adaptive filter coefficient update value Δw(m).

[0046] Therefore, not only can steady noise (such as low-frequency environmental sounds or noises) be reduced as has been possible with conventional technology, but the present disclosure can also reduce sudden noise (such as high-frequency crying or talking) to extract internal body sounds (such as low-frequency heartbeat or breathing sounds).The degree to which sudden noise (such as high-frequency crying or talking sounds) is reduced and the degree to which internal body sounds (such as low-frequency heartbeat or breathing sounds) are maintained can be quantitatively calculated.

[0047] (Processing Concept of the Improved Nonlinear Filter Unit of the Present Disclosure) The components of the improved nonlinear filter unit of the present disclosure are shown in Figure 8. The processing procedure of the improved nonlinear filter unit of the present disclosure is shown in Figure 9. The processing concept of the improved nonlinear filter unit of the present disclosure is shown in Figure 10. The improved nonlinear filter unit 5 includes a filter reduction amount calculation unit 51, a residual signal minimum smoothing unit 52, and a residual signal suppression unit 53. The improved nonlinear filter unit 5 and the internal body sound signal output unit 6 can be realized by installing the improved nonlinear filter program shown in Figure 9 in a computer.

[0048] The filter reduction amount calculation unit 51 calculates the internal body sound signal y D Frequency spectrum Y of (n) D The amount of reduction in the frequency spectrum R(k) of the residual signal r(n) for (k) is calculated, and the amount of reduction Att(k) by the improved adaptive filter unit 4 (including the filter coefficient correction unit 46) is calculated (steps S15 to S17, multiple bands b: high and low frequencies in the upper and middle rows of FIG. 10).

[0049] First, the short-term internal sound signal y D (n) amplitude or power spectrum Y D (k) is calculated (step S15). Next, the spectrum R(k) of the amplitude or power of the short-term residual signal r(n) is calculated (step S16). Next, the reduction amount Att(k)=Y D (k)-R(k), Y D (k) / R(k) or log(Y D Att(k) / R(k) is calculated (step S17). Here, the reduction amount Att(k) by the improved adaptive filter unit 4 may be an average value of a predetermined bandwidth for any of the reduction amounts in step S17.

[0050] The residual signal minimum smoothing unit 52 smooths the minimum smoothed spectrum R(k) of the residual signal r(n) so as to maintain the minimum value of the frequency spectrum R(k) of the residual signal r(n). ― (k) is calculated (step S18, multiple bands b in the upper and lower rows of FIG. 10: high frequency and low frequency).

[0051] As one method, for the frequency spectrum R(k) of the residual signal r(n), the square root of the reciprocal of the mean square of the reciprocal in a given bandwidth is taken as the minimum smoothed spectrum R(k) of the residual signal r(n). ― (k) (step S18, equation 3). i is a weighting factor given to frequency i in a given band to weaken / strengthen the smoothing in the low / high frequency range. As a separate method, the search result for the minimum value in a given bandwidth for the frequency spectrum R(k) of the residual signal r(n) is obtained as the minimum smoothed spectrum R(k) of the residual signal r(n). ― (k) (step S18).

number

[0052] The residual signal suppressor 53 reduces the complex frequency spectrum R of the residual signal r(n) as the reduction amount Att(k) by the improved adaptive filter 4 increases. C (k) is the minimum smoothed spectrum R of the residual signal r(n). ― (k), and the suppressed frequency spectrum R of the residual signal r(n) оut (k) is calculated (step S19, high frequencies in the middle and lower parts of FIG. 10).

[0053] The residual signal suppressor 53 reduces the complex frequency spectrum R of the residual signal r(n) as the reduction amount Att(k) by the improved adaptive filter 4 decreases. C (k) is the minimum smoothed spectrum R of the residual signal r(n). ― (k), and the suppressed frequency spectrum R of the residual signal r(n) оut (k) is calculated (step S19, low range in the middle and bottom rows of FIG. 10).

[0054] The residual signal suppressor 53 reduces the complex frequency spectrum R of the residual signal r(n) as the reduction amount Att(k) by the improved adaptive filter 4 decreases. C (k) is the minimum smoothed spectrum R of the residual signal r(n). ― (k), and the suppressed frequency spectrum R of the residual signal r(n) оut (k) may be used (step S19, the low frequencies in the middle and bottom rows of FIG. 10).

[0055] Specifically, the suppressed frequency spectrum R of the residual signal r(n) оut (k) is calculated based on Equation 4, where G(k) is the complex frequency spectrum R of the residual signal r(n). C (k), the suppressed frequency spectrum R of the residual signal r(n) оut S(Att(k)) is the correction gain for (k). The larger Att(k) is, the closer S(Att(k)) is to 0, and the smaller Att(k) is, the closer S(Att(k)) is to 1. Furthermore, max(a, b) is the maximum value of a and b.

number

[0056] Therefore, at frequencies where the reduction amount Att(k) by the improved adaptive filter unit 4 is large, S(Att(k))→0, 0 <R ― (k) / |R C (k)|<1, G(k)=R ― (k) / |R C (k)|, R оut (k)=R C (k)*R ― (k) / |R C On the other hand, at frequencies where the reduction amount Att(k) by the improved adaptive filter unit 4 is small, S(Att(k))→1, 0 <R ― (k) / |R C (k)|<1, G(k)→1, R оut (k) → R C (k).

[0057] The internal body sound signal output unit 6 outputs the suppressed frequency spectrum R of the residual signal r(n). оut (k) is the suppressed signal r of the residual signal r(n). оut (n) in the time domain (step S20).

[0058] Therefore, not only can steady noise (such as low-frequency environmental sounds or noises) be reduced as has been possible with conventional technology, but the present disclosure can also reduce sudden noise (such as high-frequency crying or talking) to extract internal body sounds (such as low-frequency heartbeat or breathing sounds).The degree to which sudden noise (such as high-frequency crying or talking sounds) is reduced and the degree to which internal body sounds (such as low-frequency heartbeat or breathing sounds) are maintained can be quantitatively calculated.

[0059] (Processing results of the electronic stethoscope signal processing device of the present disclosure) The processing result of the electronic stethoscope signal processing device of the present disclosure is shown in Fig. 11. In the first stage of Fig. 11, the frequency spectrum R(k) of the residual signal r(n) is calculated based on the internal sound signal y D Frequency spectrum Y of (n) D11, the reduction amount Att(k) by the improved adaptive filter unit 4 is small in the low frequency range (attenuation rate = after attenuation / before attenuation → 1) and large in the high frequency range (attenuation rate = after attenuation / before attenuation → 0).

[0060] In the third stage of Fig. 11, the minimum smoothed spectrum R of the residual signal r(n) is calculated. ― Compared with the frequency spectrum R(k) of the residual signal r(n), (k) is maintained at a minimum value in the low frequency range and at a minimum value in the high frequency range. In the fourth row of Figure 11, the complex frequency spectrum R(k) of the residual signal r(n) is C The correction gain G(k) for (k) approaches 1 at the valleys of the low and high frequencies (attenuation rate = after attenuation / before attenuation → 1) and approaches 0 at the peaks of the high frequencies (attenuation rate = after attenuation / before attenuation → 0).

[0061] In the fifth stage of Fig. 11, the suppressed frequency spectrum R of the residual signal r(n) is оut (k) is not suppressed in the low frequency range compared to the frequency spectrum R(k) of the residual signal r(n), and is the minimum smoothed spectrum R ― Towards (k), the high frequencies are suppressed.

[0062] The processing results of the conventional and presently disclosed electronic stethoscope signal processing devices are shown in Figure 12. The first row of Figure 12 shows the spectrum of the internal body sound signal before processing according to the present disclosure, clearly showing the spectrum of crying, which is a sudden external sound interference component in the frequency band above 1 kHz. The second row of Figure 12 shows the spectrum of the internal body sound signal after processing according to the present disclosure, showing that the spectrum of crying, which is a sudden external sound interference component in the frequency band above 1 kHz, has been almost completely reduced.

[0063] In this embodiment, processing is performed by the improved adaptive filter unit 4 (including the filter coefficient correction unit 46) and then by the improved nonlinear filter unit 5. As a first modification, only processing by the improved adaptive filter unit 4 (including the filter coefficient correction unit 46) may be performed. As a second modification, processing by the adaptive filter unit 41 (not including the filter coefficient correction unit 46) may be performed and then by the improved nonlinear filter unit 5. [Industrial Applicability]

[0064] The electronic stethoscope signal processing device, electronic stethoscope system, electronic stethoscope signal processing program, and electronic stethoscope signal processing method disclosed herein (1) enable medical treatment without experience by converting internal body sound signals into data, (2) enable medical treatment outside of business hours by recording internal body sound signals, and (3) enable remote medical treatment by communicating internal body sound signals. [Explanation of symbols]

[0065] S: Electronic stethoscope system E: Electronic stethoscope P: Electronic stethoscope signal processor B: Body surface 1: Extracorporeal sound collector 2: Internal sound collector 3: External and internal sound signal input section 4: Improved adaptive filter section 5: Improved nonlinear filter section 6: Internal body sound signal output section 31: Extracorporeal sound signal AD conversion unit 32: Internal body sound signal AD conversion unit 33: Internal body sound signal delay unit 41: Adaptive filter section 42: Residual signal calculation unit 43: Filter coefficient update unit 44: Provisional filter section 45: Correction gain calculation unit 46: Filter coefficient correction unit 51: Filter reduction amount calculation unit 52: Residual signal minimum smoothing unit 53: Residual signal suppression unit

Claims

1. an adaptive filter unit having adaptive filter coefficients that simulate the characteristics of extracorporeal sound passing from the outside of the electronic stethoscope to the internal sound collector, and receiving an extracorporeal sound signal obtained from the external sound collector and outputting an adaptive filter signal; a residual signal calculation unit that subtracts the adaptive filter signal from an internal sound signal obtained from the internal sound pickup device and calculates a residual signal in which the loop component of the external sound signal is reduced from the internal sound signal; a filter coefficient update unit that adds an updated value of the adaptive filter coefficient to a current value of the adaptive filter coefficient so as to reduce correlation in the time axis direction between the adaptive filter signal and the residual signal; a provisional filter unit having provisional filter coefficients set equal to the updated values ​​of the adaptive filter coefficients, receiving the extracorporeal sound signal and outputting a provisional filter signal; a correction gain calculation unit that calculates a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients to a relatively larger value for a band having a higher correlation in the frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the provisional filter signal, and calculates a correction gain for the frequency spectrum of the updated values ​​of the adaptive filter coefficients to a relatively smaller value for a band having a lower correlation in the frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the provisional filter signal; a filter coefficient correction unit that calculates a correction value for the updated value of the adaptive filter coefficient based on a correction gain for a frequency spectrum of the updated value of the adaptive filter coefficient; An electronic stethoscope signal processing device comprising:

2. The correction gain calculation unit calculates, for each band, a correction gain for the frequency spectrum of the updated value of the adaptive filter coefficients, based on a correlation in a frequency axis direction between the frequency spectrum of the residual signal and the frequency spectrum of the temporary filter signal obtained by multiplying, for each band, a correction gain for the frequency spectrum of the updated value of the adaptive filter coefficients, so as to minimize a shape error between the frequency spectra of the residual signal and the frequency spectrum of the temporary filter signal, for each band.

2. An electronic stethoscope signal processing device according to claim 1.

3. a filter reduction amount calculation unit that calculates a reduction amount of the frequency spectrum of the residual signal with respect to the frequency spectrum of the internal body sound signal, and calculates the reduction amount by the adaptive filter unit; a residual signal suppression unit that suppresses the frequency spectrum of the residual signal to a greater extent as the amount of reduction by the adaptive filter unit increases at a frequency, and maintains or suppresses the frequency spectrum of the residual signal to a lesser extent as the amount of reduction by the adaptive filter unit decreases at a frequency; 3. An electronic stethoscope signal processing device according to claim 1 or 2, further comprising:

4. an adaptive filter unit having adaptive filter coefficients that simulate the characteristics of extracorporeal sound passing from the outside of the electronic stethoscope to the internal sound collector, and receiving an extracorporeal sound signal obtained from the external sound collector and outputting an adaptive filter signal; a residual signal calculation unit that subtracts the adaptive filter signal from an internal sound signal obtained from the internal sound pickup device and calculates a residual signal in which the loop component of the external sound signal is reduced from the internal sound signal; a filter coefficient update unit that adds an updated value of the adaptive filter coefficient to a current value of the adaptive filter coefficient so as to reduce correlation in the time axis direction between the adaptive filter signal and the residual signal; a filter reduction amount calculation unit that calculates a reduction amount of the frequency spectrum of the residual signal with respect to the frequency spectrum of the internal body sound signal, and calculates the reduction amount by the adaptive filter unit; a residual signal suppression unit that suppresses the frequency spectrum of the residual signal to a greater extent as the amount of reduction by the adaptive filter unit increases at a frequency, and maintains or suppresses the frequency spectrum of the residual signal to a lesser extent as the amount of reduction by the adaptive filter unit decreases at a frequency; An electronic stethoscope signal processing device comprising:

5. a residual signal minimum smoothing unit that calculates a minimum smoothed spectrum of the residual signal so as to retain a minimum value of the frequency spectrum of the residual signal; The residual signal suppression unit suppresses the frequency spectrum of the residual signal more to a greater extent toward a minimum smoothed spectrum of the residual signal at a frequency where the amount of reduction by the adaptive filter unit is greater, while maintaining the frequency spectrum of the residual signal or suppressing it more to a lesser extent toward a minimum smoothed spectrum of the residual signal at a frequency where the amount of reduction by the adaptive filter unit is smaller.

5. An electronic stethoscope signal processing device according to claim 3 or 4.

6. An electronic stethoscope signal processing device according to any one of claims 1 to 5; the extracorporeal sound collector that collects extracorporeal sounds and outputs the extracorporeal sound signals; the internal sound collector that collects internal sounds and outputs the internal sound signals; An electronic stethoscope system comprising:

7. 6. An electronic stethoscope signal processing program for causing a computer to execute each processing step by each component of the electronic stethoscope signal processing device according to claim 1.

8. 6. An electronic stethoscope signal processing method comprising the steps of processing performed by the components of the electronic stethoscope signal processing device according to claim 1.

Images