CN109893163B - Method for automatically positioning and repairing clipping distortion waveform of electronic stethoscope - Google Patents

Abstract

The invention discloses a method for automatically positioning and repairing a clipping distortion waveform of an electronic stethoscope, which comprises the following steps: reading an auscultation signal;the signal x is differentiated to obtain a differential signal x_d(ii) a Determining K non-overlapping time intervals with clipping distortion through threshold comparison and endpoint pairing, sequencing according to the time sequence, and forming a stack in a mode of entering first and exiting later; if K is equal to or larger than 1, determining a set psi of stack interpolation data time points_K(ii) a Using a set of time points Ψ_KAnd the value x (psi) of x corresponding to each time point_K) Fitting a value in a K-th clipping distortion area through interpolation to replace the original clipping distortion value in x; removing the time interval at the top of the stack; outputting the auscultation signal x without clipping distortion. The invention can automatically detect and position the clipping distortion area of the stethoscope signal, can automatically restore the signal of the clipping distortion area and can automatically restore a plurality of clipping distortion areas with closer distance.

Description

Method for automatically positioning and repairing clipping distortion waveform of electronic stethoscope

Technical Field

The invention belongs to the technical field of stethoscopes, and relates to a method for automatically positioning and repairing a clipping distortion waveform of an electronic stethoscope.

Background

Auscultation is one of the important technologies for a physician to know the illness state of a patient in the first time, but the technology is restricted by the factors such as the visiting place, the medical skill level and the like for a long time, and various types of electronic stethoscopes are continuously emerging along with the vigorous development of the internet of things (IoT) technology, so that the possibility is brought to the real-time monitoring, automatic transcription, cloud diagnosis and intelligent diagnosis of the cardiopulmonary sound data of the patient.

The electronic stethoscope converts weak physiological sound signals (such as heart sounds, lung sounds and the like) into electric signals through a transducer (microphone), the existing electronic stethoscope is generally a piezoelectric microphone with high sensitivity and excellent frequency characteristics, and a film on a resonant cavity collects sound vibration to cause crystal deformation and finally converts the crystal deformation into current. In practical applications, however, the electronic stethoscope output current rapidly reaches saturation due to excessive compression during auscultation, so that clipping distortion occurs in the recorded auscultation signal, as in the case of the phenomenon shown in fig. 1: for example, in a pediatric consulting room, excessive pressing may occur during auscultation due to low cooperation degree of infant patients; if the patient independently uses the electronic stethoscope to upload the auscultation data to the in-process of diagnosing that carries out the high in the clouds, because improper operation also leads to appearing clipping distortion very easily, the appearance of clipping distortion makes the use of electronic stethoscope experience worsen on the one hand, and on the other hand has influenced signal quality and then has influenced follow-up heart sound location, cardiopulmonary sound automatic diagnosis's effect.

For the problem of clipping distortion of the signal caused by compression, the disadvantages of the prior art are embodied in the following aspects: a. at present, no automatic positioning method for a signal clipping distortion area exists; b. the existing automatic restoration method for the clipping distortion area of the signal is not perfect: the important premise for solving the problem is that the deletion of automatic positioning of a clipping distortion area and how to better recover a signal missing part caused by clipping according to the content of a context signal still need to be considered; c. at present, no technical scheme for repairing a plurality of clipping distortion areas with close distances exists.

In particular, for artificial intelligence prospective targets of electronic stethoscopes aiming at realizing high-precision real-time fetal monitoring, intelligent assessment of heart/lung functions, automatic diagnosis of heart/lung diseases and the like, the method can automatically locate and repair the clipping distortion in auscultation signals, and is one of important preconditions for realizing the artificial intelligence functions, so that the clipping distortion of the signals of the electronic stethoscopes is a common problem to be solved urgently by the electronic stethoscopes.

The current existing patents related to the electronic stethoscope include hardware acquisition systems, transmission systems, appearance, signal preprocessing (including noise reduction, heart sound localization, cardiopulmonary sound separation, and the like), signal intelligent analysis (fetal heart monitoring, heart sound-based heart disease intelligent diagnosis, lung sound-based respiratory disease intelligent diagnosis), and the like of the devices, but no patent provides an automatic repair solution for the problem of signal clipping distortion.

And the paper, "Emmanouilidou D, Mccolum E D, Park D E, et al, computerized Lung Sound Screening for peptide evaluation in noise Field Environments [ J ]. IEEE Transactions on biological Engineering,2018,65(7): 1564-1574" proposes a method for clipping distortion repair using cubic spline interpolation. However, the method in this document has the following problems: (1) no automatic detection and positioning method for clipping distortion area is proposed, so the method is implemented on the premise that the clipping distortion area is artificially selected firstly, and the requirements of practical application are not met; (2) the curve fitted by cubic spline interpolation used in the method can only ensure the continuity of the connection point and cannot ensure the smoothness of the connection point; the whole fitting curve is controlled by all the interpolation points, if any one of the interpolation points is changed, the whole curve is affected, which is not in accordance with the actual characteristics of auscultation data, especially for auscultation of respiratory diseases, the duration of specific respiratory extra sounds is very short (such as the duration of a single wet rale is about 20 milliseconds), and in this case, the fitting of the cubic spline interpolation brings large errors; (3) a repair strategy when two or more segments of clipping distortion areas with close distances exist is not proposed, and if the two or more segments of clipping distortion areas are directly repaired at the same time, under-fitting can be caused because the number of points to be fitted is too large.

Disclosure of Invention

The invention aims to provide a method for automatically positioning and repairing a clipping distortion waveform of an electronic stethoscope, which solves the problems that the prior art can not automatically detect and position a clipping distortion area of the stethoscope, can not automatically repair the clipping distortion area and can not repair signals of the clipping distortion area.

The purpose of the invention can be realized by the following technical scheme:

a method for automatically locating and repairing a clipped distorted waveform of an electronic stethoscope, comprising the steps of:

step 1, reading an auscultation signal sampling sequence x (N) with the duration of N in a cache, wherein N is 1, 2.

Step 2, differentiating the original signal x to obtain a differential signal x_dThe expression of the differential signal of the original signal is calculated as x_d(n) ═ x (n) -x (n-1), where x is_d(1)＝0；

Step 3, according to the differential signal x_dAfter threshold comparison and end point pairing, K non-overlapping time intervals (K is more than or equal to 0) with clipping distortion are determined, and the time intervals sequentially comprise: [ n ] of_K,begin,n_K,end],[n_K-1,begin,n_K-1,end],...,[n_1,begin,n_1,end]Forming a stack according to a mode that a clipping distortion interval appearing later enters first and then exits;

step 4, if K is equal to 0, the section of data has no clipping distortion area and does not need to be repaired, the step 7 is directly entered, and if K is larger than or equal to 1, the earliest clipping distortion area [ n ] in the stack is determined_K,begin,n_K,end]Interpolation data time point set psi required for repairing_K；

Step 5, utilizing the time point set psi_KAnd the value x (psi) of x corresponding to each time point_K) Fitting a value in a K-th section clipping distortion area through Hermite interpolation to replace the original clipping distortion value in x, and updating x;

step 6, removing the time interval [ n ] at the top of the stack_K,begin,n_K,end]Let K be K-1 and return to step 4.

And 7, outputting the auscultation signal x without clipping distortion.

Further, the forming of the stack in the step 3 specifically includes the following steps:

s1, finding out differential signal x_dThe time point corresponding to the numerical value of which the middle absolute value is greater than the threshold value alpha.max (| x |) forms a set { n |₁,n₂,...,n_LWherein, alpha is more than 0 and less than 1, which is a preset constant;

s2, from the set n₁,n₂,...,n_LFind adjacent x in }_dPairing every two time points with opposite signs of the corresponding values, and determining the time points as the starting points n of a certain clipping distortion interval_k,beginAnd an end point n_k,end(ii) a If a single unpaired time point is found near the beginning or the end of the whole data, the data is additionally paired with the previous section or the next section of data;

s3, according to the pairwise matching results, if K non-overlapped time intervals are obtained, the time is [ n ] respectively_K,begin,n_K,end],[n_K-1,begin,n_K-1,end],...,[n_1,begin,n_1,end]In the order of (2), the stack is formed in such a way that the clipping distortion interval appearing later first enters the stack.

Further, the set Ψ of interpolated data time points required in step 4_KThe acquisition method specifically comprises the following steps:

h1 from time n_K,beginStarting from n, finding n against the direction of the time axis_K,beginTime point n corresponding to the 3 rd zero crossing point on the first x_K,left3If the corresponding point on x is not equal to 0, the time point corresponding to the value closest to 0 is taken;

h2 from time n_K,endStarting from n, finding n along the time axis_K,endTime point n corresponding to the 1 st zero crossing point on the starting x_K,right1If the corresponding point on x is not exactly equal to 0, the time point corresponding to the value closest to 0 is taken;

h3, determining and repairing clipping distortion area [ n_K,begin,n_K,end]Set of required data time points Ψ_KIs [ n ]_K,left3,n_K,begin-P]∪[n_K,end+P,n_K,right1]And P is a reserved point set by considering the conversion time of the stethoscope output to the saturated state when excessive pressing is performed, and the value of P is generally a non-negative integer of 1-10.

Further, the method for updating x in step 5 specifically includes the following steps:

f1, if time set Ψ_KHas a common T_k+1 time points: n is₀,n₁,…,

Estimate the data point x (n) at each time point_i) Derivative x' (n)_i),i＝0,1,...,T_k；

F2, any point x (m) needing fitting repair, wherein m∈[n_K,begin-P+1,n_K,end+P-1]Is a point in time on the region to be fitted, using Ψ_KMiddle T_kData point x (n) at +1 time points_i) And its derivative x' (n)_i),i＝0,1,...,T_kAnd obtaining Hermite interpolation:

f3, updating the value of the K-th interval to be repaired in x: x (m) ═ x_H(m),m∈[n_K,begin-P+1,n_K,end+P-1]。

The invention has the beneficial effects that:

the method for automatically positioning and repairing the clipping distortion waveform of the electronic stethoscope realizes automatic detection and positioning of the clipping distortion area of the stethoscope signal by differentiating the signal, comparing the signal with a threshold value and pairing; selecting an interpolation data time point set required for restoration by combining automatic positioning of a clipping distortion area with a zero crossing point, and automatically restoring a signal of the clipping distortion area through Hermite interpolation; by adopting the 'first-in and last-out' characteristic of the stack, the automatic repair of a plurality of clipping distortion areas can be sequentially realized, the repair problem when the clipping distortion areas are very close to each other can be solved, and the method has the characteristics of high positioning and repairing speed and high accuracy.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.

FIG. 1 is a diagram illustrating clipping distortion of an auscultation signal in the prior art;

FIG. 2 is a schematic diagram of a method of automatically locating and repairing a clipped distortion waveform of an electronic stethoscope according to the present invention;

FIG. 3 is a schematic view of repair in example 1 of the present invention;

FIG. 4 is a schematic view of repair in example 2 of the present invention;

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

As shown in fig. 1, a method for automatically locating and repairing a clipped distorted waveform of an electronic stethoscope has the following steps:

step 1, reading an auscultation signal sampling sequence x (N) with the duration of N in a cache, wherein N is 1, 2.

Step 2, differentiating the original signal x to obtain a differential signal x_dThe expression of the differential signal of the original signal is calculated as x_d(n) ═ x (n) -x (n-1), where x is_d(1)＝0；

Step 3, according to the differential signal x_dAfter threshold comparison and end point pairing, K non-overlapping time intervals (K is more than or equal to 0) with clipping distortion are determined, and the time intervals sequentially comprise: [ n ] of_K,begin,n_K,end],[n_K-1,begin,n_K-1,end],...,[n_1,begin,n_1,end]Forming a stack according to a mode that a clipping distortion interval appearing later enters first and then exits;

forming a stack in step 3, specifically comprising the steps of:

s1, finding out differential signal x_dThe time point corresponding to the numerical value of which the middle absolute value is greater than the threshold value alpha.max (| x |) forms a set { n |₁,n₂,...,n_LWherein, alpha is more than 0 and less than 1, which is a preset constant;

s2, from the set n₁,n₂,...,n_LFind adjacent x in }_dThe time points of the upper corresponding values with opposite signs advance pairwiseLine pairing is carried out, and the line pairing is determined to be respectively a starting point n of a certain clipping distortion interval_k,beginAnd an end point n_k,end(ii) a If a single unpaired time point is found near the beginning or the end of the whole data, the data is additionally paired with the previous section or the next section of data;

s3, according to the pairwise matching results, if K non-overlapped time intervals are obtained, the time is [ n ] respectively_K,begin,n_K,end],[n_K-1,begin,n_K-1,end],...,[n_1,begin,n_1,end]In the order of (2), the stack is formed in such a way that the clipping distortion interval appearing later first enters the stack.

Step 4, if K is equal to 0, the section of data has no clipping distortion area and does not need to be repaired, the step 7 is directly entered, and if K is larger than or equal to 1, the earliest clipping distortion area [ n ] in the stack is determined_K,begin,n_K,end]Interpolation data time point set psi required for repairing_K；

In step 4, if K is more than or equal to 1, determining the earliest clipping distortion region [ n ] in the stack_K,begin,n_K,end]Interpolation data time point set psi required for repairing_KThe acquisition method specifically comprises the following steps:

h1 from time n_K,beginStarting from n, finding n against the direction of the time axis_K,beginTime point n corresponding to the 3 rd zero crossing point on the first x_K,left3If the corresponding point on x is not equal to 0, the time point corresponding to the value closest to 0 is taken;

h2 from time n_K,endStarting from n, finding n along the time axis_K,endTime point n corresponding to the 1 st zero crossing point on the starting x_K,right1If the corresponding point on x is not exactly equal to 0, the time point corresponding to the value closest to 0 is taken;

h3, determining and repairing clipping distortion area [ n_K,begin,n_K,end]Set of required data time points Ψ_KIs [ n ]_K,left3,n_K,begin-P]∪[n_K,end+P,n_K,right1]Where P is the transition of the stethoscope output to saturation when excessive compression is taken into accountThe reserved points set by the time change are generally non-negative integers of 1-10.

Step 5, utilizing the time point set psi_KAnd the value x (psi) of x corresponding to each time point_K) Fitting a value in a K-th section clipping distortion area through Hermite interpolation to replace the original clipping distortion value in x, and updating x;

a method of updating x, comprising the steps of:

f1, if time set Ψ_KHas a common T_k+1 time points: n is₀,n₁,…,

Estimate the data point x (n) at each time point_i) Derivative x' (n)_i),i＝0,1,...,T_k；

F2, any point x (m) needing fitting repair, wherein m is epsilon [ n ]_K,begin-P+1,n_K,end+P-1]Is a point in time on the region to be fitted, using Ψ_KMiddle T_k+1 data points x (n)_i) And its derivative x' (n)_i),i＝0,1,...,T_kAnd obtaining Hermite interpolation:

f3, updating the value of the K-th interval to be repaired in x: x (m) ═ x_H(m),m∈[n_K,begin-P+1,n_K,end+P-1]。

Step 6, removing the time interval [ n ] at the top of the stack_K,begin,n_K,end]Let K be K-1 and return to step 4.

And 7, outputting the auscultation signal x without clipping distortion.

Example 1:

reading lung sound auscultation data x of a section of pediatric pneumonia patient, wherein the lung sound auscultation data x comprises two clipping distortion areas, and automatically positioning and repairing the clipping distortion areas, wherein the data sampling rate is 4KHz, and the time duration is 2.5 seconds.

First, for the segmentData differencing, difference signal x_dAs shown in fig. 3(a), it can be seen that there are two peaks with opposite polarities and large amplitude at the beginning and the end of the clipping distortion region.

Then, a differential signal x is found_dFinding out the adjacent x from the time point corresponding to the numerical value of which the medium absolute value is greater than the threshold value 0.8 multiplied by max (| x |)_dPairing the time points with opposite signs of the corresponding values, detecting two clipping distortion areas, and determining the respective starting point and end point, as shown in fig. 3 (b);

finally, the two located clipping distortion intervals are placed into a stack according to a reverse time sequence, an interpolation point set of the clipping distortion intervals is sequentially determined by using stack operation, the clipping distortion intervals are sequentially repaired through Hermite interpolation, and the signal updating is completed, and finally the repaired signal is as shown in fig. 3(c), wherein a reserved point P which is set by considering the conversion time of the stethoscope output to the saturation state when the stethoscope is excessively pressed is 5.

Example 2:

reading a section of normal human heart sound auscultation data x, wherein the data comprises 5 clipping distortion areas, wherein two clipping distortion areas are close to each other and are separated from each other by 0.0136 seconds, automatically positioning and repairing the clipping distortion areas, the data sampling rate is 4KHz, and the time duration is 1.5 seconds.

First, the segment of data is differentiated to obtain a differential signal x_dAs shown in fig. 4(a), it can be seen that at 5, both the beginning and the end of the clipping distortion region have a pair of peaks of opposite polarity and large amplitude.

Then, a differential signal x is found_dFinding out the adjacent x from the time point corresponding to the numerical value of which the medium absolute value is greater than the threshold value 0.8 multiplied by max (| x |)_dPairing the time points with opposite signs of the corresponding values, detecting all 5 clipping distortion areas, and determining respective starting points and end points of the clipping distortion areas, as shown in fig. 4 (b);

finally, the 5 located clipping distortion intervals are placed into a stack according to a reverse time sequence, an interpolation point set of the 5 located clipping distortion intervals is sequentially determined by using stack operation, the signals are sequentially repaired through Hermite interpolation, and the finally repaired signals are shown as a figure 4(c), wherein a reserved point P set by considering the conversion time of the stethoscope output converted into a saturation state when the stethoscope is excessively pressed is 5, and even if two clipping distortion areas which are very close to each other are seen, the method provided by the invention can still realize more accurate repair.

The foregoing is merely exemplary and illustrative of the principles of the present invention and various modifications, additions and substitutions of the specific embodiments described herein may be made by those skilled in the art without departing from the principles of the present invention or exceeding the scope of the claims set forth herein.

Claims (3)

1. A method for automatically locating and repairing a clipping distortion waveform of an electronic stethoscope is characterized in that: comprises the following steps:

step 1, reading an auscultation signal sampling sequence x (N) with the duration of N in a cache, wherein N is 1, 2.

Step 2, differentiating the original signal x to obtain a differential signal x_dThe expression of the differential signal of the original signal is calculated as x_d(n) ═ x (n) -x (n-1), where x is_d(1)＝0；

Step 3, according to the differential signal x_dAfter threshold comparison and end point pairing, K non-overlapping time intervals (K is more than or equal to 0) with clipping distortion are determined, and the time intervals sequentially comprise: [ n ] of_K,begin,n_K,end],[n_K-1,begin,n_K-1,end],...,[n_1,begin,n_1,end]Forming a stack according to a mode that a clipping distortion interval appearing later enters first and then exits;

step 4, if K is equal to 0, the section of data has no clipping distortion area and does not need to be repaired, the step 7 is directly entered, and if K is larger than or equal to 1, the earliest clipping distortion area [ n ] in the stack is determined_K,begin,n_K,end]Interpolation data time point set psi required for repairing_K；

Step 5, utilizing the time point set psi_KAnd the value x (psi) of x corresponding to each time point_K) Fitting a value in a K-th section clipping distortion area through Hermite interpolation to replace the original clipping distortion value in x, and updating x;

step 6, removing the time interval [ n ] at the top of the stack_K,begin,n_K,end]Making K equal to K-1, and returning to the step 4;

step 7, outputting the auscultation signal without clipping distortion;

the step 3 of forming a stack specifically includes the following steps:

s1, finding out differential signal x_dThe time point corresponding to the numerical value of which the middle absolute value is greater than the threshold value alpha.max (| x |) forms a set { n |₁,n₂,...,n_LWherein, alpha is more than 0 and less than 1, which is a preset constant;

s2, from the set n₁,n₂,...,n_LFind adjacent x in }_dPairing every two time points with opposite signs of the corresponding values, and determining the time points as the starting points n of a certain clipping distortion interval_k,beginAnd an end point n_k,end(ii) a If a single unpaired time point is found near the beginning or the end of the whole data, the data is additionally paired with the previous section or the next section of data;

s3, according to the pairwise matching results, if K non-overlapped time intervals are obtained, the time is [ n ] respectively_K,begin,n_K,end],[n_K-1,begin,n_K-1,end],...,[n_1,begin,n_1,end]In the order of (2), the stack is formed in such a way that the clipping distortion interval appearing later first enters the stack.

2. The method of claim 1, wherein the method comprises the steps of: the interpolation data time point set Ψ required in the step 4_KThe acquisition method specifically comprises the following steps:

h1 from time n_K,beginStarting from n, finding n against the direction of the time axis_K,beginTime point n corresponding to the 3 rd zero crossing point on the first x_K,left3If the corresponding point on x is not equal to 0, then the closest point is takenThe time point corresponding to the value of 0;

h2 from time n_K,endStarting from n, finding n along the time axis_K,endTime point n corresponding to the 1 st zero crossing point on the starting x_K,right1If the corresponding point on x is not exactly equal to 0, the time point corresponding to the value closest to 0 is taken;

h3, determining and repairing clipping distortion area [ n_K,begin,n_K,end]Set of required data time points Ψ_KIs [ n ]_K,left3,n_K,begin-P]∪[n_K,end+P,n_K,right1]And P is a reserved point set by considering the conversion time of the stethoscope output to the saturated state when excessive pressing is performed, and the value of P is generally a non-negative integer of 1-10.

3. The method of claim 1, wherein the method comprises the steps of: the method for updating x in step 5 specifically includes the following steps:

f1, if time set Ψ_KHas a common T_k+1 time points:

estimate the data point x (n) at each time point_i) Derivative x' (n)_i),i＝0,1,...,T_k；

F2, any point x (m) needing fitting repair, wherein m is epsilon [ n ]_K,begin-P+1,n_K,end+P-1]Is a point in time on the region to be fitted, using Ψ_KMiddle T_kData point x (n) at +1 time points_i) And its derivative x' (n)_i),i＝0,1,...,T_kAnd obtaining Hermite interpolation:

；

f3, updating the value of the K-th interval to be repaired in x: x (m) ═ x_H(m),m∈[n_K,begin-P+1,n_K,end+P-1]。

Images